fix: merge repo_update — add archived+template, keep default_branch+confirm from main

Adds a repo_update tool exposing PATCH /api/v1/repos/{owner}/{name}
with optional pointer fields (archived, description, private,
website, template). Only fields set by the caller are sent on the
wire, so the server patches exactly what was asked for.

Originally needed to archive ingestion-svc cleanly instead of
leaving a README tombstone, and to flip template-go-{agent,web}
to template=true so create_project_from_template stops failing
the "is not marked as template" guard.

Wire-level enforcement of "at least one field" returns ErrValidation
before any network call, preventing no-op PATCHes.

private=false (making a repo public) is allowed but flagged in the
tool description with a "verify intent before calling" warning.
The earlier issue draft suggested an ntfy confirmation hook for
that path — out of scope for this PR; the warning string is the
minimum that fits inside the tool surface today.

Wires NewRepoUpdate into cmd/gitea-mcp/main.go alongside the rest
of the repo_* family.

Closes #12

.aider.conf.yml

@@ -0,0 +1,2 @@
+read: .aider.conventions.md
+auto-commits: false

.aider.conventions.md

@@ -0,0 +1,250 @@
+# Agent context — Mathias workspace
+
+<!-- Canonical root context for all AI coding agents.
+     Lives at: ~/dev/.context/AGENT.md
+     Applies to every project under ~/dev/ unless overridden.
+     
+     Run `task context:sync` from ~/dev/ to regenerate harness-specific files.
+     Project-level context in .context/PROJECT.md layers on top of this. -->
+
+## Who I am
+
+I'm Mathias, a digital product manager and technology consultant based in Sweden.
+I build software, research emerging tech, and deliver consulting engagements
+for clients under NDA. I work across AI/ML, financial automation, web applications,
+and climate/sustainability tech.
+
+## How I work with agents
+
+- I think like a product manager — I care about *why* before *how*
+- I want agents to be opinionated and push back, not just execute blindly
+- I prefer concise responses; skip ceremony and get to the point
+- When I say "build this", I mean production-quality with tests, not a demo
+- Ask me before making irreversible changes or adding heavy dependencies
+- I work with confidential client data — never send it to cloud APIs unless I explicitly say it's OK
+
+## Behavior rules
+
+These rules apply to every task across every project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Default stack
+
+| Layer | Default | Fallback | Last resort |
+|-------|---------|----------|-------------|
+| Language | Go | Python | TypeScript, Java, C |
+| UI | HTMX + Templ | Server-rendered HTML | React (only if SPA is justified) |
+| Build | Task (taskfile.dev) | Make | — |
+| Containers | Docker Compose (dev), k3s (prod) | — | — |
+| DB | PostgreSQL + sqlc | SQLite | — |
+| Search | pgvector (vector), BM25 | Qdrant (when >1M vectors or hybrid retrieval) | — |
+| Logging | slog (structured) | — | — |
+| Testing | Table-driven, testify | — | — |
+
+Exploratory: Rust, Zig — I'll tell you when I want these.
+
+## Code conventions
+
+- **Go style**: golines, gofumpt, golangci-lint
+- **Errors**: `fmt.Errorf("operation: %w", err)` — never naked, never log-and-return
+- **Naming**: stdlib conventions, no stuttering
+- **Architecture**: prefer stdlib over frameworks, constructor injection, env-var config parsed into typed structs
+- **Git**: conventional commits (`feat:`, `fix:`, `chore:`), one concern per PR, PR describes *why* not *what*
+- **Security**: no secrets in code, govulncheck before adding deps, SOPS for encrypted config
+- **Dependencies**: prefer stdlib. testify, slog, templ, sqlc are pre-approved; anything else needs justification in the commit message
+
+## Infrastructure
+
+Three machines on Tailscale:
+
+| Machine | Role | Key specs |
+|---------|------|-----------|
+| koala | GPU inference, heavy compute | RTX 5070, runs k3s + llama-swap + shared postgres18/pgvector |
+| iguana | Services, builds | M2 Ultra Mac |
+| flamingo | Daily driver, edge | Mac mini, ~/dev is here |
+
+- **Model routing**: LiteLLM in front of llama-swap (local) + cloud APIs (when permitted)
+- **Orchestration**: k3s cluster across all three machines
+- **Networking**: Tailscale mesh
+
+## Project landscape
+
+All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
+
+Organized in thematic folders:
+
+| Folder | Focus | Count |
+|--------|-------|-------|
+| `GO/` | Go web frameworks, API integrations, learning projects | ~10 |
+| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 |
+| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 |
+| `QKX/` | Invoice processing, financial automation, payment systems | ~13 |
+| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
+
+See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project.
+
+### Key active projects
+
+- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP
+- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
+- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
+- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
+- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
+
+## Knowledge base
+
+When available, agents can query the shared knowledge base:
+
+- **MCP**: `mcp://hyperguild.<TAILNET>.ts.net:3100/knowledge`
+- **HTTP**: `http://hyperguild.<TAILNET>.ts.net:3100/api/v1/search`
+
+<!-- TODO: replace <TAILNET> placeholder with the real Tailscale tailnet
+     name once hyperguild is deployed. Until then, agents that try to
+     reach the knowledge service on a host where it isn't running will
+     get DNS NXDOMAIN, which is the desired fail-loudly behavior. -->
+- **Scoping**: defaults to `public` collection; client projects filter to `{client}` + `public`
+
+## Client work rules
+
+When working on a project tagged with a client name:
+1. Never send code, data, or context to cloud APIs — use local models only
+2. Never reference other client projects or their data
+3. Keep all artifacts within the client's git org / directory
+4. Treat everything as confidential unless told otherwise
+
+## Harness-agnostic principles
+
+This context is designed to work with any AI coding tool:
+- Claude Code, Cursor, Aider, Open WebUI, Charmbracelet Mods/Crush
+- Pi Coding Agent, Mistral Vibe, Antigravity
+- Any tool that accepts a system prompt or reads a markdown context file
+
+The canonical source is always `.context/AGENT.md` (root) and `.context/PROJECT.md` (per-project).
+Derived files are committed (see *How context propagates* below) so a `git pull` on any host yields full agent context with no setup.
+
+## How context propagates
+
+Canonical sources of truth:
+- Universal: `~/dev/.context/AGENT.md` (this file)
+- Project: `<repo>/.context/PROJECT.md` (per-repo)
+
+Derived files (committed, regenerated by `task context:sync`):
+- `CLAUDE.md`, `AGENTS.md`, `.cursorrules`, `.aider.conventions.md`,
+  `.context/system-prompt.txt`
+
+Workflow:
+1. Edit a canonical file. Run `task context:sync`. Commit canonical and
+   derived together. Push.
+2. On any other host, `git pull` brings both. Claude Code (tree-walking)
+   uses `CLAUDE.md`; Crush / Pi / Antigravity (cwd-only) use `AGENTS.md`;
+   Cursor uses `.cursorrules`; Aider uses `.aider.conventions.md`.
+3. `task check` runs `context:sync` then asserts `git status --porcelain`
+   is empty over the derived files (catches both modified-tracked drift
+   and missing-untracked adapters). A drift fails the check with a
+   message telling you to stage the regenerated files.
+
+Behavior rules in this file and per-project rules in `PROJECT.md` apply
+unconditionally on every host, every harness.
+
+## Engineering Skills
+
+Shared engineering skills are available in `~/dev/.skills/`. Load on demand via the index.
+
+See `~/dev/.skills/SKILLS_INDEX.md` for the full list with descriptions and "use when" triggers.
+
+Key skills:
+- **TDD**: always write tests first — load `tdd` skill
+- **Code Review**: load `code-review` skill before any review
+- **SOLID/Clean Code**: load `solid` or `clean-code` skill for design work
+- **Problem first**: load `problem-analysis` skill before coding non-trivial features
+
+---
+
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM

.context/PROJECT.md

@@ -0,0 +1,79 @@
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM

.context/mcp.json

@@ -0,0 +1,22 @@
+{
+  "mcpServers": {
+    "brain": {
+      "type": "http",
+      "url": "https://brain-mcp.d-ma.be/mcp",
+      "headers": {
+        "Authorization": "Bearer ${BRAIN_MCP_TOKEN}"
+      }
+    },
+    "gitea": {
+      "type": "http",
+      "url": "https://git-mcp.d-ma.be/mcp",
+      "headers": {
+        "Authorization": "Bearer ${GITEA_MCP_TOKEN}"
+      }
+    },
+    "infra": {
+      "type": "http",
+      "url": "https://infra-mcp.d-ma.be/mcp"
+    }
+  }
+}

.context/system-prompt.txt

@@ -0,0 +1,257 @@
+You are a coding assistant working on a specific project.
+Follow all conventions from both the root agent context and project context.
+
+---
+
+# Agent context — Mathias workspace
+
+<!-- Canonical root context for all AI coding agents.
+     Lives at: ~/dev/.context/AGENT.md
+     Applies to every project under ~/dev/ unless overridden.
+     
+     Run `task context:sync` from ~/dev/ to regenerate harness-specific files.
+     Project-level context in .context/PROJECT.md layers on top of this. -->
+
+## Who I am
+
+I'm Mathias, a digital product manager and technology consultant based in Sweden.
+I build software, research emerging tech, and deliver consulting engagements
+for clients under NDA. I work across AI/ML, financial automation, web applications,
+and climate/sustainability tech.
+
+## How I work with agents
+
+- I think like a product manager — I care about *why* before *how*
+- I want agents to be opinionated and push back, not just execute blindly
+- I prefer concise responses; skip ceremony and get to the point
+- When I say "build this", I mean production-quality with tests, not a demo
+- Ask me before making irreversible changes or adding heavy dependencies
+- I work with confidential client data — never send it to cloud APIs unless I explicitly say it's OK
+
+## Behavior rules
+
+These rules apply to every task across every project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Default stack
+
+| Layer | Default | Fallback | Last resort |
+|-------|---------|----------|-------------|
+| Language | Go | Python | TypeScript, Java, C |
+| UI | HTMX + Templ | Server-rendered HTML | React (only if SPA is justified) |
+| Build | Task (taskfile.dev) | Make | — |
+| Containers | Docker Compose (dev), k3s (prod) | — | — |
+| DB | PostgreSQL + sqlc | SQLite | — |
+| Search | pgvector (vector), BM25 | Qdrant (when >1M vectors or hybrid retrieval) | — |
+| Logging | slog (structured) | — | — |
+| Testing | Table-driven, testify | — | — |
+
+Exploratory: Rust, Zig — I'll tell you when I want these.
+
+## Code conventions
+
+- **Go style**: golines, gofumpt, golangci-lint
+- **Errors**: `fmt.Errorf("operation: %w", err)` — never naked, never log-and-return
+- **Naming**: stdlib conventions, no stuttering
+- **Architecture**: prefer stdlib over frameworks, constructor injection, env-var config parsed into typed structs
+- **Git**: conventional commits (`feat:`, `fix:`, `chore:`), one concern per PR, PR describes *why* not *what*
+- **Security**: no secrets in code, govulncheck before adding deps, SOPS for encrypted config
+- **Dependencies**: prefer stdlib. testify, slog, templ, sqlc are pre-approved; anything else needs justification in the commit message
+
+## Infrastructure
+
+Three machines on Tailscale:
+
+| Machine | Role | Key specs |
+|---------|------|-----------|
+| koala | GPU inference, heavy compute | RTX 5070, runs k3s + llama-swap + shared postgres18/pgvector |
+| iguana | Services, builds | M2 Ultra Mac |
+| flamingo | Daily driver, edge | Mac mini, ~/dev is here |
+
+- **Model routing**: LiteLLM in front of llama-swap (local) + cloud APIs (when permitted)
+- **Orchestration**: k3s cluster across all three machines
+- **Networking**: Tailscale mesh
+
+## Project landscape
+
+All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
+
+Organized in thematic folders:
+
+| Folder | Focus | Count |
+|--------|-------|-------|
+| `GO/` | Go web frameworks, API integrations, learning projects | ~10 |
+| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 |
+| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 |
+| `QKX/` | Invoice processing, financial automation, payment systems | ~13 |
+| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
+
+See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project.
+
+### Key active projects
+
+- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP
+- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
+- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
+- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
+- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
+
+## Knowledge base
+
+When available, agents can query the shared knowledge base:
+
+- **MCP**: `mcp://hyperguild.<TAILNET>.ts.net:3100/knowledge`
+- **HTTP**: `http://hyperguild.<TAILNET>.ts.net:3100/api/v1/search`
+
+<!-- TODO: replace <TAILNET> placeholder with the real Tailscale tailnet
+     name once hyperguild is deployed. Until then, agents that try to
+     reach the knowledge service on a host where it isn't running will
+     get DNS NXDOMAIN, which is the desired fail-loudly behavior. -->
+- **Scoping**: defaults to `public` collection; client projects filter to `{client}` + `public`
+
+## Client work rules
+
+When working on a project tagged with a client name:
+1. Never send code, data, or context to cloud APIs — use local models only
+2. Never reference other client projects or their data
+3. Keep all artifacts within the client's git org / directory
+4. Treat everything as confidential unless told otherwise
+
+## Harness-agnostic principles
+
+This context is designed to work with any AI coding tool:
+- Claude Code, Cursor, Aider, Open WebUI, Charmbracelet Mods/Crush
+- Pi Coding Agent, Mistral Vibe, Antigravity
+- Any tool that accepts a system prompt or reads a markdown context file
+
+The canonical source is always `.context/AGENT.md` (root) and `.context/PROJECT.md` (per-project).
+Derived files are committed (see *How context propagates* below) so a `git pull` on any host yields full agent context with no setup.
+
+## How context propagates
+
+Canonical sources of truth:
+- Universal: `~/dev/.context/AGENT.md` (this file)
+- Project: `<repo>/.context/PROJECT.md` (per-repo)
+
+Derived files (committed, regenerated by `task context:sync`):
+- `CLAUDE.md`, `AGENTS.md`, `.cursorrules`, `.aider.conventions.md`,
+  `.context/system-prompt.txt`
+
+Workflow:
+1. Edit a canonical file. Run `task context:sync`. Commit canonical and
+   derived together. Push.
+2. On any other host, `git pull` brings both. Claude Code (tree-walking)
+   uses `CLAUDE.md`; Crush / Pi / Antigravity (cwd-only) use `AGENTS.md`;
+   Cursor uses `.cursorrules`; Aider uses `.aider.conventions.md`.
+3. `task check` runs `context:sync` then asserts `git status --porcelain`
+   is empty over the derived files (catches both modified-tracked drift
+   and missing-untracked adapters). A drift fails the check with a
+   message telling you to stage the regenerated files.
+
+Behavior rules in this file and per-project rules in `PROJECT.md` apply
+unconditionally on every host, every harness.
+
+## Engineering Skills
+
+Shared engineering skills are available in `~/dev/.skills/`. Load on demand via the index.
+
+See `~/dev/.skills/SKILLS_INDEX.md` for the full list with descriptions and "use when" triggers.
+
+Key skills:
+- **TDD**: always write tests first — load `tdd` skill
+- **Code Review**: load `code-review` skill before any review
+- **SOLID/Clean Code**: load `solid` or `clean-code` skill for design work
+- **Problem first**: load `problem-analysis` skill before coding non-trivial features
+
+---
+
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM
+
+---

.cursorrules

@@ -0,0 +1,253 @@
+# Cursor rules — auto-generated
+# Do not edit. Run: task context:sync
+
+# Agent context — Mathias workspace
+
+<!-- Canonical root context for all AI coding agents.
+     Lives at: ~/dev/.context/AGENT.md
+     Applies to every project under ~/dev/ unless overridden.
+     
+     Run `task context:sync` from ~/dev/ to regenerate harness-specific files.
+     Project-level context in .context/PROJECT.md layers on top of this. -->
+
+## Who I am
+
+I'm Mathias, a digital product manager and technology consultant based in Sweden.
+I build software, research emerging tech, and deliver consulting engagements
+for clients under NDA. I work across AI/ML, financial automation, web applications,
+and climate/sustainability tech.
+
+## How I work with agents
+
+- I think like a product manager — I care about *why* before *how*
+- I want agents to be opinionated and push back, not just execute blindly
+- I prefer concise responses; skip ceremony and get to the point
+- When I say "build this", I mean production-quality with tests, not a demo
+- Ask me before making irreversible changes or adding heavy dependencies
+- I work with confidential client data — never send it to cloud APIs unless I explicitly say it's OK
+
+## Behavior rules
+
+These rules apply to every task across every project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Default stack
+
+| Layer | Default | Fallback | Last resort |
+|-------|---------|----------|-------------|
+| Language | Go | Python | TypeScript, Java, C |
+| UI | HTMX + Templ | Server-rendered HTML | React (only if SPA is justified) |
+| Build | Task (taskfile.dev) | Make | — |
+| Containers | Docker Compose (dev), k3s (prod) | — | — |
+| DB | PostgreSQL + sqlc | SQLite | — |
+| Search | pgvector (vector), BM25 | Qdrant (when >1M vectors or hybrid retrieval) | — |
+| Logging | slog (structured) | — | — |
+| Testing | Table-driven, testify | — | — |
+
+Exploratory: Rust, Zig — I'll tell you when I want these.
+
+## Code conventions
+
+- **Go style**: golines, gofumpt, golangci-lint
+- **Errors**: `fmt.Errorf("operation: %w", err)` — never naked, never log-and-return
+- **Naming**: stdlib conventions, no stuttering
+- **Architecture**: prefer stdlib over frameworks, constructor injection, env-var config parsed into typed structs
+- **Git**: conventional commits (`feat:`, `fix:`, `chore:`), one concern per PR, PR describes *why* not *what*
+- **Security**: no secrets in code, govulncheck before adding deps, SOPS for encrypted config
+- **Dependencies**: prefer stdlib. testify, slog, templ, sqlc are pre-approved; anything else needs justification in the commit message
+
+## Infrastructure
+
+Three machines on Tailscale:
+
+| Machine | Role | Key specs |
+|---------|------|-----------|
+| koala | GPU inference, heavy compute | RTX 5070, runs k3s + llama-swap + shared postgres18/pgvector |
+| iguana | Services, builds | M2 Ultra Mac |
+| flamingo | Daily driver, edge | Mac mini, ~/dev is here |
+
+- **Model routing**: LiteLLM in front of llama-swap (local) + cloud APIs (when permitted)
+- **Orchestration**: k3s cluster across all three machines
+- **Networking**: Tailscale mesh
+
+## Project landscape
+
+All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
+
+Organized in thematic folders:
+
+| Folder | Focus | Count |
+|--------|-------|-------|
+| `GO/` | Go web frameworks, API integrations, learning projects | ~10 |
+| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 |
+| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 |
+| `QKX/` | Invoice processing, financial automation, payment systems | ~13 |
+| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
+
+See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project.
+
+### Key active projects
+
+- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP
+- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
+- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
+- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
+- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
+
+## Knowledge base
+
+When available, agents can query the shared knowledge base:
+
+- **MCP**: `mcp://hyperguild.<TAILNET>.ts.net:3100/knowledge`
+- **HTTP**: `http://hyperguild.<TAILNET>.ts.net:3100/api/v1/search`
+
+<!-- TODO: replace <TAILNET> placeholder with the real Tailscale tailnet
+     name once hyperguild is deployed. Until then, agents that try to
+     reach the knowledge service on a host where it isn't running will
+     get DNS NXDOMAIN, which is the desired fail-loudly behavior. -->
+- **Scoping**: defaults to `public` collection; client projects filter to `{client}` + `public`
+
+## Client work rules
+
+When working on a project tagged with a client name:
+1. Never send code, data, or context to cloud APIs — use local models only
+2. Never reference other client projects or their data
+3. Keep all artifacts within the client's git org / directory
+4. Treat everything as confidential unless told otherwise
+
+## Harness-agnostic principles
+
+This context is designed to work with any AI coding tool:
+- Claude Code, Cursor, Aider, Open WebUI, Charmbracelet Mods/Crush
+- Pi Coding Agent, Mistral Vibe, Antigravity
+- Any tool that accepts a system prompt or reads a markdown context file
+
+The canonical source is always `.context/AGENT.md` (root) and `.context/PROJECT.md` (per-project).
+Derived files are committed (see *How context propagates* below) so a `git pull` on any host yields full agent context with no setup.
+
+## How context propagates
+
+Canonical sources of truth:
+- Universal: `~/dev/.context/AGENT.md` (this file)
+- Project: `<repo>/.context/PROJECT.md` (per-repo)
+
+Derived files (committed, regenerated by `task context:sync`):
+- `CLAUDE.md`, `AGENTS.md`, `.cursorrules`, `.aider.conventions.md`,
+  `.context/system-prompt.txt`
+
+Workflow:
+1. Edit a canonical file. Run `task context:sync`. Commit canonical and
+   derived together. Push.
+2. On any other host, `git pull` brings both. Claude Code (tree-walking)
+   uses `CLAUDE.md`; Crush / Pi / Antigravity (cwd-only) use `AGENTS.md`;
+   Cursor uses `.cursorrules`; Aider uses `.aider.conventions.md`.
+3. `task check` runs `context:sync` then asserts `git status --porcelain`
+   is empty over the derived files (catches both modified-tracked drift
+   and missing-untracked adapters). A drift fails the check with a
+   message telling you to stage the regenerated files.
+
+Behavior rules in this file and per-project rules in `PROJECT.md` apply
+unconditionally on every host, every harness.
+
+## Engineering Skills
+
+Shared engineering skills are available in `~/dev/.skills/`. Load on demand via the index.
+
+See `~/dev/.skills/SKILLS_INDEX.md` for the full list with descriptions and "use when" triggers.
+
+Key skills:
+- **TDD**: always write tests first — load `tdd` skill
+- **Code Review**: load `code-review` skill before any review
+- **SOLID/Clean Code**: load `solid` or `clean-code` skill for design work
+- **Problem first**: load `problem-analysis` skill before coding non-trivial features
+
+---
+
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM

.skills/go-patterns/SKILL.md

@@ -0,0 +1,42 @@
+---
+name: go-patterns
+description: Go project patterns — endpoint checklist, error handling, HTMX responses, dependency policy. Use when writing Go code, adding endpoints, or reviewing Go PRs.
+---
+
+# Go project patterns
+
+## New endpoint checklist
+1. Define request/response types in `types.go`
+2. Write handler in `handlers.go` using `http.HandlerFunc`
+3. Add route in `routes.go`
+4. Write table-driven test in `handlers_test.go`
+5. Run `task check` before committing
+
+## Error handling pattern
+```go
+if err != nil {
+    return fmt.Errorf("descriptiveOperation: %w", err)
+}
+```
+Never log and return — do one or the other.
+
+## HTMX response pattern
+```go
+func (h *Handler) ListItems(w http.ResponseWriter, r *http.Request) {
+    items, err := h.store.List(r.Context())
+    if err != nil {
+        http.Error(w, "failed to list items", http.StatusInternalServerError)
+        return
+    }
+    if r.Header.Get("HX-Request") == "true" {
+        h.templates.Render(w, "items/_list", items)
+        return
+    }
+    h.templates.Render(w, "items/index", items)
+}
+```
+
+## Dependency policy
+- Prefer stdlib: `net/http`, `encoding/json`, `database/sql`
+- Allowed without justification: `testify`, `slog`, `templ`, `sqlc`
+- Needs justification in commit message: anything else

.skills/htmx-patterns/SKILL.md

@@ -0,0 +1,31 @@
+---
+name: htmx-patterns
+description: HTMX conventions — default attributes, form patterns, validation errors, hypermedia-first API design. Use when writing HTMX templates or Go handlers that return HTML fragments.
+---
+
+# HTMX patterns
+
+## Default attributes
+Always include on interactive elements:
+- `hx-indicator` for loading states
+- `hx-swap="innerHTML"` as default (explicit over implicit)
+- `hx-target` pointing to a specific ID, never `this` in production
+
+## Form pattern
+```html
+<form hx-post="/items" hx-target="#item-list" hx-swap="beforeend" hx-indicator="#spinner">
+  <input type="text" name="title" required>
+  <button type="submit">Add</button>
+  <span id="spinner" class="htmx-indicator">...</span>
+</form>
+```
+
+## Server-sent validation errors
+Return 422 with the error fragment, swap into the form's error container:
+```html
+hx-target-422="#form-errors"
+```
+
+## Prefer hypermedia over JSON
+If the endpoint returns data for display, return an HTML fragment.
+Only use JSON for machine-to-machine APIs or when a non-browser client needs it.

AGENTS.md

@@ -0,0 +1,250 @@
+# Agent context — Mathias workspace
+
+<!-- Canonical root context for all AI coding agents.
+     Lives at: ~/dev/.context/AGENT.md
+     Applies to every project under ~/dev/ unless overridden.
+     
+     Run `task context:sync` from ~/dev/ to regenerate harness-specific files.
+     Project-level context in .context/PROJECT.md layers on top of this. -->
+
+## Who I am
+
+I'm Mathias, a digital product manager and technology consultant based in Sweden.
+I build software, research emerging tech, and deliver consulting engagements
+for clients under NDA. I work across AI/ML, financial automation, web applications,
+and climate/sustainability tech.
+
+## How I work with agents
+
+- I think like a product manager — I care about *why* before *how*
+- I want agents to be opinionated and push back, not just execute blindly
+- I prefer concise responses; skip ceremony and get to the point
+- When I say "build this", I mean production-quality with tests, not a demo
+- Ask me before making irreversible changes or adding heavy dependencies
+- I work with confidential client data — never send it to cloud APIs unless I explicitly say it's OK
+
+## Behavior rules
+
+These rules apply to every task across every project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Default stack
+
+| Layer | Default | Fallback | Last resort |
+|-------|---------|----------|-------------|
+| Language | Go | Python | TypeScript, Java, C |
+| UI | HTMX + Templ | Server-rendered HTML | React (only if SPA is justified) |
+| Build | Task (taskfile.dev) | Make | — |
+| Containers | Docker Compose (dev), k3s (prod) | — | — |
+| DB | PostgreSQL + sqlc | SQLite | — |
+| Search | pgvector (vector), BM25 | Qdrant (when >1M vectors or hybrid retrieval) | — |
+| Logging | slog (structured) | — | — |
+| Testing | Table-driven, testify | — | — |
+
+Exploratory: Rust, Zig — I'll tell you when I want these.
+
+## Code conventions
+
+- **Go style**: golines, gofumpt, golangci-lint
+- **Errors**: `fmt.Errorf("operation: %w", err)` — never naked, never log-and-return
+- **Naming**: stdlib conventions, no stuttering
+- **Architecture**: prefer stdlib over frameworks, constructor injection, env-var config parsed into typed structs
+- **Git**: conventional commits (`feat:`, `fix:`, `chore:`), one concern per PR, PR describes *why* not *what*
+- **Security**: no secrets in code, govulncheck before adding deps, SOPS for encrypted config
+- **Dependencies**: prefer stdlib. testify, slog, templ, sqlc are pre-approved; anything else needs justification in the commit message
+
+## Infrastructure
+
+Three machines on Tailscale:
+
+| Machine | Role | Key specs |
+|---------|------|-----------|
+| koala | GPU inference, heavy compute | RTX 5070, runs k3s + llama-swap + shared postgres18/pgvector |
+| iguana | Services, builds | M2 Ultra Mac |
+| flamingo | Daily driver, edge | Mac mini, ~/dev is here |
+
+- **Model routing**: LiteLLM in front of llama-swap (local) + cloud APIs (when permitted)
+- **Orchestration**: k3s cluster across all three machines
+- **Networking**: Tailscale mesh
+
+## Project landscape
+
+All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
+
+Organized in thematic folders:
+
+| Folder | Focus | Count |
+|--------|-------|-------|
+| `GO/` | Go web frameworks, API integrations, learning projects | ~10 |
+| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 |
+| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 |
+| `QKX/` | Invoice processing, financial automation, payment systems | ~13 |
+| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
+
+See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project.
+
+### Key active projects
+
+- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP
+- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
+- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
+- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
+- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
+
+## Knowledge base
+
+When available, agents can query the shared knowledge base:
+
+- **MCP**: `mcp://hyperguild.<TAILNET>.ts.net:3100/knowledge`
+- **HTTP**: `http://hyperguild.<TAILNET>.ts.net:3100/api/v1/search`
+
+<!-- TODO: replace <TAILNET> placeholder with the real Tailscale tailnet
+     name once hyperguild is deployed. Until then, agents that try to
+     reach the knowledge service on a host where it isn't running will
+     get DNS NXDOMAIN, which is the desired fail-loudly behavior. -->
+- **Scoping**: defaults to `public` collection; client projects filter to `{client}` + `public`
+
+## Client work rules
+
+When working on a project tagged with a client name:
+1. Never send code, data, or context to cloud APIs — use local models only
+2. Never reference other client projects or their data
+3. Keep all artifacts within the client's git org / directory
+4. Treat everything as confidential unless told otherwise
+
+## Harness-agnostic principles
+
+This context is designed to work with any AI coding tool:
+- Claude Code, Cursor, Aider, Open WebUI, Charmbracelet Mods/Crush
+- Pi Coding Agent, Mistral Vibe, Antigravity
+- Any tool that accepts a system prompt or reads a markdown context file
+
+The canonical source is always `.context/AGENT.md` (root) and `.context/PROJECT.md` (per-project).
+Derived files are committed (see *How context propagates* below) so a `git pull` on any host yields full agent context with no setup.
+
+## How context propagates
+
+Canonical sources of truth:
+- Universal: `~/dev/.context/AGENT.md` (this file)
+- Project: `<repo>/.context/PROJECT.md` (per-repo)
+
+Derived files (committed, regenerated by `task context:sync`):
+- `CLAUDE.md`, `AGENTS.md`, `.cursorrules`, `.aider.conventions.md`,
+  `.context/system-prompt.txt`
+
+Workflow:
+1. Edit a canonical file. Run `task context:sync`. Commit canonical and
+   derived together. Push.
+2. On any other host, `git pull` brings both. Claude Code (tree-walking)
+   uses `CLAUDE.md`; Crush / Pi / Antigravity (cwd-only) use `AGENTS.md`;
+   Cursor uses `.cursorrules`; Aider uses `.aider.conventions.md`.
+3. `task check` runs `context:sync` then asserts `git status --porcelain`
+   is empty over the derived files (catches both modified-tracked drift
+   and missing-untracked adapters). A drift fails the check with a
+   message telling you to stage the regenerated files.
+
+Behavior rules in this file and per-project rules in `PROJECT.md` apply
+unconditionally on every host, every harness.
+
+## Engineering Skills
+
+Shared engineering skills are available in `~/dev/.skills/`. Load on demand via the index.
+
+See `~/dev/.skills/SKILLS_INDEX.md` for the full list with descriptions and "use when" triggers.
+
+Key skills:
+- **TDD**: always write tests first — load `tdd` skill
+- **Code Review**: load `code-review` skill before any review
+- **SOLID/Clean Code**: load `solid` or `clean-code` skill for design work
+- **Problem first**: load `problem-analysis` skill before coding non-trivial features
+
+---
+
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM

CLAUDE.md

@@ -0,0 +1,79 @@
+# Project context
+
+<!-- Canonical project context. Edit this, run `task context:sync`.
+     Root agent context from ~/dev/.context/AGENT.md is automatically
+     prepended for harnesses that don't walk the directory tree. -->
+
+## Identity
+
+- **Name**: gitea-mcp
+- **Owner**: Mathias
+- **Client**: personal
+- **Repo**: https://gitea.d-ma.be/mathias/gitea-mcp
+- **Status**: active
+
+## Stack
+
+- **Primary language**: Go
+- **UI layer**: HTMX + Templ (when applicable)
+- **Fallback languages**: Python, TypeScript (justify in PR if used)
+- **Build**: Task (taskfile.dev), not Make
+- **Containers**: Docker (compose for dev, k3s for deploy)
+- **Target infra**: koala (GPU workloads), iguana (services), flamingo (edge)
+
+## Conventions
+
+### Code style
+- Go: follow `golines`, `gofumpt`, `golangci-lint` with project config
+- Tests: table-driven, in `_test.go` next to source, `testify` for assertions
+- Errors: wrap with `fmt.Errorf("operation: %w", err)`, no naked returns
+- Naming: stdlib conventions, no stuttering (`http.Client` not `http.HTTPClient`)
+
+### Architecture preferences
+- Prefer standard library over frameworks (net/http over gin/echo)
+- Dependency injection via constructor functions, not containers
+- Configuration via environment variables, parsed at startup into a typed struct
+- Structured logging via `slog`
+
+### Git
+- Conventional commits: `feat:`, `fix:`, `chore:`, `docs:`, `refactor:`
+- Branch naming: `feat/short-description`, `fix/short-description`
+- PRs: one concern per PR, description explains *why* not *what*
+
+### Security
+- No secrets in code, ever — use env vars or SOPS-encrypted files
+- Client data never leaves local network unless explicitly cleared
+- Dependencies: audit with `govulncheck` before adding
+
+## Knowledge base access
+
+This project can query the shared knowledge base via MCP or HTTP:
+
+- **MCP endpoint**: `mcp://localhost:3100/knowledge`
+- **HTTP fallback**: `http://localhost:3100/api/v1/search`
+- **Scoping**: queries are filtered to collection `personal` + `public`
+
+## Behavior rules
+
+These rules apply to every task in this project, regardless of harness.
+
+1. **No assumptions.** Don't hide confusion — surface it. Surface tradeoffs explicitly.
+   Think before coding; if the problem is unclear, ask or state assumptions before acting.
+2. **Minimum viable code.** Solve with the smallest change that works. Nothing
+   speculative, no "while we're here" cleanups, no premature abstractions. Simplicity first.
+3. **Surgical changes.** Touch only what the task requires. Leave unrelated code,
+   files, and formatting alone. Diffs should be small and reviewable.
+4. **Goal-driven execution.** Define clear success criteria up front for every task.
+   Loop — implement, verify, refine — until those criteria are met. Don't claim
+   completion without evidence (tests pass, command output, observed behavior).
+
+## Agent instructions
+
+When acting as a coding agent on this project:
+
+1. Read this file and all `SKILL.md` files in `.skills/` before starting work
+2. Run `task check` before committing (lint + test + vet)
+3. If unsure about a convention, check `DECISIONS.md` or ask
+4. Never modify files outside the project root without explicit permission
+5. When adding a dependency, explain why in the commit message
+6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM

Taskfile.yml

@@ -14,9 +14,42 @@ tasks:
   lint:
     desc: Run golangci-lint
     cmds: [golangci-lint run ./...]
-  check:
+  vet:
-    desc: Lint, vet, and test (used by CI)
     cmds:
-      - golangci-lint run ./...
       - go vet ./...
-      - go test ./... -race -count=1
+      - govulncheck ./... || true
+
+  check:
+    desc: Run all checks (context freshness + lint + test + vet)
+    cmds:
+      - cmd: |
+          if [ -n "${CI:-}" ]; then
+            echo "✓ context sync: skipped in CI"
+          else
+            bash scripts/context-sync.sh
+            drift=$(git status --porcelain -- AGENTS.md CLAUDE.md .cursorrules .aider.conventions.md .context/system-prompt.txt 2>/dev/null)
+            if [ -n "$drift" ]; then
+              echo "ERROR: derived adapters drifted from canonical context." >&2
+              echo "$drift" >&2
+              echo "" >&2
+              echo "Run: git add AGENTS.md CLAUDE.md .cursorrules .aider.conventions.md .context/system-prompt.txt" >&2
+              echo "     git commit -m 'chore: re-sync context adapters'" >&2
+              exit 1
+            fi
+            echo "✓ context: canonical and adapters are in sync"
+          fi
+      - task: lint
+      - task: test
+      - task: vet
+
+  context:sync:
+    desc: Regenerate all harness-specific context files
+    cmds:
+      - bash scripts/context-sync.sh
+
+  context:sync:claude:
+    cmds: [bash scripts/context-sync.sh claude]
+  context:sync:agents:
+    cmds: [bash scripts/context-sync.sh agents]
+  context:sync:cursor:
+    cmds: [bash scripts/context-sync.sh cursor]

cmd/gitea-mcp/main.go

@@ -1,6 +1,8 @@
 package main
 
 import (
+	"context"
+	"encoding/json"
 	"log/slog"
 	"net/http"
 	"os"
@@ -23,7 +25,14 @@ func main() {
 		os.Exit(1)
 	}
 
-	giteaClient := gitea.NewClient(cfg.GiteaBaseURL, "")
+	ctx := context.Background()
+
+	jwtValidator, err := auth.NewJWTValidator(ctx, cfg.DexIssuerURL, cfg.MCPAudience)
+	if err != nil {
+		logger.Warn("jwt validator init failed; JWT auth disabled", "err", err)
+	}
+
+	giteaClient := gitea.NewClient(cfg.GiteaBaseURL, cfg.DefaultToken)
 	ownerAllow := allowlist.New(cfg.AllowedOwners)
 
 	reg := registry.New()
@@ -31,6 +40,7 @@ func main() {
 	reg.Register(tools.NewRepoGet(giteaClient, ownerAllow))
 	reg.Register(tools.NewRepoSearch(giteaClient, ownerAllow))
 	reg.Register(tools.NewRepoStatus(giteaClient, ownerAllow))
+	reg.Register(tools.NewRepoUpdate(giteaClient, ownerAllow))
 	reg.Register(tools.NewFileRead(giteaClient, ownerAllow))
 	reg.Register(tools.NewFileWriteBranch(giteaClient, ownerAllow))
 	reg.Register(tools.NewFileDelete(giteaClient, ownerAllow))
@@ -59,7 +69,7 @@ func main() {
 
 	mux := http.NewServeMux()
 	mux.Handle("/mcp", mcp.OriginAllowlist(cfg.OriginAllowlist)(
-		auth.BearerMiddleware(cfg.GiteaBaseURL,
+		auth.BearerMiddleware(jwtValidator, cfg.StaticToken,
 			auth.CallerMiddleware(mcpSrv),
 		),
 	))
@@ -73,11 +83,14 @@ func main() {
 			return
 		}
 		w.Header().Set("Content-Type", "application/json")
-		w.WriteHeader(http.StatusOK)
+		payload := map[string]any{
-		_, _ = w.Write([]byte(`{"authorization_servers":[]}`))
+			"resource":             cfg.MCPResourceURL,
-	})
+			"authorization_servers": []string{},
-	mux.HandleFunc("/.well-known/oauth-authorization-server", func(w http.ResponseWriter, r *http.Request) {
+		}
-		http.NotFound(w, r)
+		if cfg.DexIssuerURL != "" {
+			payload["authorization_servers"] = []string{cfg.DexIssuerURL}
+		}
+		_ = json.NewEncoder(w).Encode(payload)
 	})
 
 	addr := ":" + cfg.Port

go.mod

@@ -2,10 +2,24 @@ module gitea.d-ma.be/mathias/gitea-mcp
 
 go 1.26.2
 
+require (
+	github.com/hashicorp/golang-lru/v2 v2.0.7
+	github.com/lestrrat-go/jwx/v2 v2.1.6
+	github.com/stretchr/testify v1.11.1
+)
+
 require (
 	github.com/davecgh/go-spew v1.1.1 // indirect
-	github.com/hashicorp/golang-lru/v2 v2.0.7 // indirect
+	github.com/decred/dcrd/dcrec/secp256k1/v4 v4.4.0 // indirect
+	github.com/goccy/go-json v0.10.3 // indirect
+	github.com/lestrrat-go/blackmagic v1.0.3 // indirect
+	github.com/lestrrat-go/httpcc v1.0.1 // indirect
+	github.com/lestrrat-go/httprc v1.0.6 // indirect
+	github.com/lestrrat-go/iter v1.0.2 // indirect
+	github.com/lestrrat-go/option v1.0.1 // indirect
 	github.com/pmezard/go-difflib v1.0.0 // indirect
-	github.com/stretchr/testify v1.11.1 // indirect
+	github.com/segmentio/asm v1.2.0 // indirect
+	golang.org/x/crypto v0.32.0 // indirect
+	golang.org/x/sys v0.31.0 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,11 +1,39 @@
+github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/decred/dcrd/dcrec/secp256k1/v4 v4.4.0 h1:NMZiJj8QnKe1LgsbDayM4UoHwbvwDRwnI3hwNaAHRnc=
+github.com/decred/dcrd/dcrec/secp256k1/v4 v4.4.0/go.mod h1:ZXNYxsqcloTdSy/rNShjYzMhyjf0LaoftYK0p+A3h40=
+github.com/goccy/go-json v0.10.3 h1:KZ5WoDbxAIgm2HNbYckL0se1fHD6rz5j4ywS6ebzDqA=
+github.com/goccy/go-json v0.10.3/go.mod h1:oq7eo15ShAhp70Anwd5lgX2pLfOS3QCiwU/PULtXL6M=
 github.com/hashicorp/golang-lru/v2 v2.0.7 h1:a+bsQ5rvGLjzHuww6tVxozPZFVghXaHOwFs4luLUK2k=
 github.com/hashicorp/golang-lru/v2 v2.0.7/go.mod h1:QeFd9opnmA6QUJc5vARoKUSoFhyfM2/ZepoAG6RGpeM=
+github.com/lestrrat-go/blackmagic v1.0.3 h1:94HXkVLxkZO9vJI/w2u1T0DAoprShFd13xtnSINtDWs=
+github.com/lestrrat-go/blackmagic v1.0.3/go.mod h1:6AWFyKNNj0zEXQYfTMPfZrAXUWUfTIZ5ECEUEJaijtw=
+github.com/lestrrat-go/httpcc v1.0.1 h1:ydWCStUeJLkpYyjLDHihupbn2tYmZ7m22BGkcvZZrIE=
+github.com/lestrrat-go/httpcc v1.0.1/go.mod h1:qiltp3Mt56+55GPVCbTdM9MlqhvzyuL6W/NMDA8vA5E=
+github.com/lestrrat-go/httprc v1.0.6 h1:qgmgIRhpvBqexMJjA/PmwSvhNk679oqD1RbovdCGW8k=
+github.com/lestrrat-go/httprc v1.0.6/go.mod h1:mwwz3JMTPBjHUkkDv/IGJ39aALInZLrhBp0X7KGUZlo=
+github.com/lestrrat-go/iter v1.0.2 h1:gMXo1q4c2pHmC3dn8LzRhJfP1ceCbgSiT9lUydIzltI=
+github.com/lestrrat-go/iter v1.0.2/go.mod h1:Momfcq3AnRlRjI5b5O8/G5/BvpzrhoFTZcn06fEOPt4=
+github.com/lestrrat-go/jwx/v2 v2.1.6 h1:hxM1gfDILk/l5ylers6BX/Eq1m/pnxe9NBwW6lVfecA=
+github.com/lestrrat-go/jwx/v2 v2.1.6/go.mod h1:Y722kU5r/8mV7fYDifjug0r8FK8mZdw0K0GpJw/l8pU=
+github.com/lestrrat-go/option v1.0.1 h1:oAzP2fvZGQKWkvHa1/SAcFolBEca1oN+mQ7eooNBEYU=
+github.com/lestrrat-go/option v1.0.1/go.mod h1:5ZHFbivi4xwXxhxY9XHDe2FHo6/Z7WWmtT7T5nBBp3I=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/segmentio/asm v1.2.0 h1:9BQrFxC+YOHJlTlHGkTrFWf59nbL3XnCoFLTwDCI7ys=
+github.com/segmentio/asm v1.2.0/go.mod h1:BqMnlJP91P8d+4ibuonYZw9mfnzI9HfxselHZr5aAcs=
+github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/testify v1.6.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
+github.com/stretchr/testify v1.7.1/go.mod h1:6Fq8oRcR53rry900zMqJjRRixrwX3KX962/h/Wwjteg=
 github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
 github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+golang.org/x/crypto v0.32.0 h1:euUpcYgM8WcP71gNpTqQCn6rC2t6ULUPiOzfWaXVVfc=
+golang.org/x/crypto v0.32.0/go.mod h1:ZnnJkOaASj8g0AjIduWNlq2NRxL0PlBrbKVyZ6V/Ugc=
+golang.org/x/sys v0.31.0 h1:ioabZlmFYtWhL+TRYpcnNlLwhyxaM9kWTDEmfnprqik=
+golang.org/x/sys v0.31.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/yaml.v3 v3.0.0-20200313102051-9f266ea9e77c/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=

internal/auth/bearer.go

@@ -1,49 +1,42 @@
 package auth
 
 import (
-	"context"
+	"crypto/subtle"
 	"net/http"
 	"strings"
-	"time"
 )
 
-type tokenKey struct{}
+// BearerMiddleware authenticates requests via the Authorization header.
-
+//
-// BearerMiddleware validates the incoming bearer token as a Gitea PAT by
+// A request is allowed when:
-// calling GET /api/v1/user. The validated token is stored in context for
+//
-// downstream use by the Gitea client.
+//  1. The Bearer token is a valid JWT issued by the configured Dex OIDC server, or
-func BearerMiddleware(giteaBaseURL string, next http.Handler) http.Handler {
+//  2. The Bearer token matches staticToken (constant-time compare).
-	hc := &http.Client{Timeout: 5 * time.Second}
+//
+// Any other case — including missing or empty Authorization header — returns 401.
+//
+// The Gitea service PAT is intentionally NOT used to authenticate the caller:
+// it is only used by the Gitea client for upstream API calls. Decoupling the
+// two prevents the MCP endpoint from being reachable anonymously when a service
+// PAT happens to be configured.
+func BearerMiddleware(jwtValidator *JWTValidator, staticToken string, next http.Handler) http.Handler {
 	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		token, ok := strings.CutPrefix(r.Header.Get("Authorization"), "Bearer ")
+		bearer, hasBearer := strings.CutPrefix(r.Header.Get("Authorization"), "Bearer ")
-		if !ok || token == "" {
+		if !hasBearer || bearer == "" {
 			http.Error(w, "unauthorized", http.StatusUnauthorized)
 			return
 		}
-		req, err := http.NewRequestWithContext(r.Context(), http.MethodGet, giteaBaseURL+"/api/v1/user", nil)
+
-		if err != nil {
+		if jwtValidator.Validate(r.Context(), bearer) {
-			http.Error(w, "unauthorized", http.StatusUnauthorized)
+			next.ServeHTTP(w, r)
 			return
 		}
-		req.Header.Set("Authorization", "token "+token)
+
-		resp, err := hc.Do(req)
+		if staticToken != "" && subtle.ConstantTimeCompare([]byte(bearer), []byte(staticToken)) == 1 {
-		if err != nil || resp.StatusCode != http.StatusOK {
+			next.ServeHTTP(w, r)
-			if resp != nil {
-				_ = resp.Body.Close()
-			}
-			http.Error(w, "unauthorized", http.StatusUnauthorized)
 			return
 		}
-		_ = resp.Body.Close()
+
-		ctx := context.WithValue(r.Context(), tokenKey{}, token)
+		http.Error(w, "unauthorized", http.StatusUnauthorized)
-		next.ServeHTTP(w, r.WithContext(ctx))
 	})
 }
-
-// TokenFromContext returns the validated Gitea PAT stored by BearerMiddleware.
-func TokenFromContext(ctx context.Context) string {
-	if v, ok := ctx.Value(tokenKey{}).(string); ok {
-		return v
-	}
-	return ""
-}

internal/auth/bearer_test.go

@@ -10,73 +10,83 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-func TestBearerMiddleware_NoAuthHeader(t *testing.T) {
+func okHandler(called *bool) http.Handler {
-	srv := httptest.NewServer(auth.BearerMiddleware("https://gitea.example.com",
+	return http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
-		http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		if called != nil {
+			*called = true
+		}
 		w.WriteHeader(http.StatusOK)
-		}),
+	})
-	))
+}
+
+func TestBearerMiddleware_NoAuthHeader(t *testing.T) {
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, "", okHandler(nil)))
 	defer srv.Close()
 
 	resp, err := http.Post(srv.URL+"/mcp", "application/json", nil)
 	require.NoError(t, err)
-	defer resp.Body.Close()
+	defer func() { _ = resp.Body.Close() }()
 	assert.Equal(t, http.StatusUnauthorized, resp.StatusCode)
 }
 
-func TestBearerMiddleware_InvalidToken(t *testing.T) {
+func TestBearerMiddleware_NoAuthHeader_RejectsEvenWhenStaticConfigured(t *testing.T) {
-	// Mock Gitea that rejects the token
+	// A configured staticToken must not allow unauthenticated callers through.
-	giteaMock := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, "any-static", okHandler(nil)))
-		w.WriteHeader(http.StatusUnauthorized)
+	defer srv.Close()
-	}))
-	defer giteaMock.Close()
 
-	srv := httptest.NewServer(auth.BearerMiddleware(giteaMock.URL,
+	resp, err := http.Post(srv.URL+"/mcp", "application/json", nil)
-		http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+	require.NoError(t, err)
-			w.WriteHeader(http.StatusOK)
+	defer func() { _ = resp.Body.Close() }()
-		}),
+	assert.Equal(t, http.StatusUnauthorized, resp.StatusCode)
-	))
+}
+
+func TestBearerMiddleware_EmptyBearer(t *testing.T) {
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, "static", okHandler(nil)))
 	defer srv.Close()
 
 	req, _ := http.NewRequest(http.MethodPost, srv.URL+"/mcp", nil)
-	req.Header.Set("Authorization", "Bearer bad-token")
+	req.Header.Set("Authorization", "Bearer ")
 	resp, err := http.DefaultClient.Do(req)
 	require.NoError(t, err)
-	defer resp.Body.Close()
+	defer func() { _ = resp.Body.Close() }()
 	assert.Equal(t, http.StatusUnauthorized, resp.StatusCode)
 }
 
-func TestBearerMiddleware_ValidToken(t *testing.T) {
+func TestBearerMiddleware_StaticToken_Valid(t *testing.T) {
-	const token = "valid-pat"
+	const staticToken = "my-static-token"
-
-	// Mock Gitea that accepts the token and returns a user
-	giteaMock := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		assert.Equal(t, "token "+token, r.Header.Get("Authorization"))
-		w.WriteHeader(http.StatusOK)
-	}))
-	defer giteaMock.Close()
-
 	called := false
-	srv := httptest.NewServer(auth.BearerMiddleware(giteaMock.URL,
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, staticToken, okHandler(&called)))
-		http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-			called = true
-			// Token must be available in context for downstream Gitea client
-			assert.Equal(t, token, auth.TokenFromContext(r.Context()))
-			w.WriteHeader(http.StatusOK)
-		}),
-	))
 	defer srv.Close()
 
 	req, _ := http.NewRequest(http.MethodPost, srv.URL+"/mcp", nil)
-	req.Header.Set("Authorization", "Bearer "+token)
+	req.Header.Set("Authorization", "Bearer "+staticToken)
 	resp, err := http.DefaultClient.Do(req)
 	require.NoError(t, err)
-	defer resp.Body.Close()
+	defer func() { _ = resp.Body.Close() }()
 	assert.Equal(t, http.StatusOK, resp.StatusCode)
 	assert.True(t, called)
 }
 
-func TestTokenFromContext_Empty(t *testing.T) {
+func TestBearerMiddleware_StaticToken_Invalid(t *testing.T) {
-	req := httptest.NewRequest(http.MethodGet, "/", nil)
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, "correct-token", okHandler(nil)))
-	assert.Equal(t, "", auth.TokenFromContext(req.Context()))
+	defer srv.Close()
+
+	req, _ := http.NewRequest(http.MethodPost, srv.URL+"/mcp", nil)
+	req.Header.Set("Authorization", "Bearer wrong-token")
+	resp, err := http.DefaultClient.Do(req)
+	require.NoError(t, err)
+	defer func() { _ = resp.Body.Close() }()
+	assert.Equal(t, http.StatusUnauthorized, resp.StatusCode)
+}
+
+func TestBearerMiddleware_UnknownBearer_NoStatic_NoJWT(t *testing.T) {
+	srv := httptest.NewServer(auth.BearerMiddleware(nil, "", okHandler(nil)))
+	defer srv.Close()
+
+	req, _ := http.NewRequest(http.MethodPost, srv.URL+"/mcp", nil)
+	req.Header.Set("Authorization", "Bearer random-unknown-token")
+	resp, err := http.DefaultClient.Do(req)
+	require.NoError(t, err)
+	defer func() { _ = resp.Body.Close() }()
+	assert.Equal(t, http.StatusUnauthorized, resp.StatusCode)
 }

internal/auth/jwt.go

@@ -0,0 +1,79 @@
+package auth
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"time"
+
+	"github.com/lestrrat-go/jwx/v2/jwk"
+	"github.com/lestrrat-go/jwx/v2/jwt"
+)
+
+// JWTValidator validates bearer tokens as JWTs issued by a Dex OIDC server.
+// A nil JWTValidator always returns false — JWT validation is disabled.
+type JWTValidator struct {
+	issuer  string
+	aud     string
+	cache   *jwk.Cache
+	jwksURI string
+}
+
+// NewJWTValidator creates a validator by fetching the OIDC discovery document
+// from issuerURL. Returns nil, nil when issuerURL is empty (disabled).
+func NewJWTValidator(ctx context.Context, issuerURL, audience string) (*JWTValidator, error) {
+	if issuerURL == "" {
+		return nil, nil
+	}
+
+	resp, err := http.Get(issuerURL + "/.well-known/openid-configuration")
+	if err != nil {
+		return nil, fmt.Errorf("fetch oidc discovery: %w", err)
+	}
+	defer func() { _ = resp.Body.Close() }()
+
+	var doc struct {
+		JWKSURI string `json:"jwks_uri"`
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&doc); err != nil {
+		return nil, fmt.Errorf("decode oidc discovery: %w", err)
+	}
+
+	cache := jwk.NewCache(ctx)
+	if err := cache.Register(doc.JWKSURI, jwk.WithRefreshInterval(time.Hour)); err != nil {
+		return nil, fmt.Errorf("register jwks uri: %w", err)
+	}
+	// warm the cache immediately so first request doesn't block
+	if _, err := cache.Refresh(ctx, doc.JWKSURI); err != nil {
+		return nil, fmt.Errorf("warm jwks cache: %w", err)
+	}
+
+	return &JWTValidator{
+		issuer:  issuerURL,
+		aud:     audience,
+		cache:   cache,
+		jwksURI: doc.JWKSURI,
+	}, nil
+}
+
+// Validate returns true if rawToken is a valid JWT signed by the OIDC server.
+func (v *JWTValidator) Validate(ctx context.Context, rawToken string) bool {
+	if v == nil {
+		return false
+	}
+	keySet, err := v.cache.Get(ctx, v.jwksURI)
+	if err != nil {
+		return false
+	}
+	opts := []jwt.ParseOption{
+		jwt.WithKeySet(keySet),
+		jwt.WithIssuer(v.issuer),
+		jwt.WithValidate(true),
+	}
+	if v.aud != "" {
+		opts = append(opts, jwt.WithAudience(v.aud))
+	}
+	_, err = jwt.Parse([]byte(rawToken), opts...)
+	return err == nil
+}

internal/config/config.go

@@ -8,16 +8,26 @@ import (
 type Config struct {
 	Port            string   // GITEA_MCP_PORT, default 8080
 	GiteaBaseURL    string   // GITEA_BASE_URL, e.g. https://gitea.d-ma.be
+	DefaultToken    string   // GITEA_MCP_DEFAULT_TOKEN, service PAT; used by Gitea client for all upstream calls
+	StaticToken     string   // GITEA_MCP_STATIC_TOKEN, optional static bearer for service-to-service auth
 	AllowedOwners   []string // GITEA_MCP_ALLOWED_OWNERS, comma-separated, default "mathias"
 	OriginAllowlist []string // GITEA_MCP_ORIGIN_ALLOWLIST, comma-separated
+	DexIssuerURL    string   // DEX_ISSUER_URL, e.g. https://auth.d-ma.be; empty disables JWT auth
+	MCPAudience     string   // MCP_AUDIENCE, JWT audience claim to validate, e.g. claude-ai
+	MCPResourceURL  string   // MCP_RESOURCE_URL, this server's public URL for /.well-known metadata
 }
 
 func Load() (Config, error) {
 	cfg := Config{
 		Port:            envOr("GITEA_MCP_PORT", "8080"),
 		GiteaBaseURL:    os.Getenv("GITEA_BASE_URL"),
+		DefaultToken:    os.Getenv("GITEA_MCP_DEFAULT_TOKEN"),
+		StaticToken:     os.Getenv("GITEA_MCP_STATIC_TOKEN"),
 		AllowedOwners:   splitCSV(envOr("GITEA_MCP_ALLOWED_OWNERS", "mathias")),
 		OriginAllowlist: splitCSV(os.Getenv("GITEA_MCP_ORIGIN_ALLOWLIST")),
+		DexIssuerURL:    os.Getenv("DEX_ISSUER_URL"),
+		MCPAudience:     os.Getenv("MCP_AUDIENCE"),
+		MCPResourceURL:  os.Getenv("MCP_RESOURCE_URL"),
 	}
 	return cfg, nil
 }

internal/gitea/client.go

@@ -7,7 +7,6 @@ import (
 	"net/http"
 	"time"
 
-	"gitea.d-ma.be/mathias/gitea-mcp/internal/auth"
 	"github.com/hashicorp/golang-lru/v2/expirable"
 )
 
@@ -50,10 +49,7 @@ func (c *Client) doOnce(ctx context.Context, method, path string, body []byte) (
 	if err != nil {
 		return nil, 0, err
 	}
-	token := auth.TokenFromContext(ctx)
+	token := c.token
-	if token == "" {
-		token = c.token
-	}
 	if token != "" {
 		req.Header.Set("Authorization", "token "+token)
 	}
@@ -119,10 +115,7 @@ func (c *Client) doRaw(ctx context.Context, method, path string, body []byte) (*
 	if err != nil {
 		return nil, err
 	}
-	token := auth.TokenFromContext(ctx)
+	token := c.token
-	if token == "" {
-		token = c.token
-	}
 	if token != "" {
 		req.Header.Set("Authorization", "token "+token)
 	}

internal/gitea/repos.go

@@ -86,3 +86,34 @@ func (c *Client) GetRepo(ctx context.Context, owner, name string) (*Repo, error)
 	}
 	return &r, nil
 }
+
+// EditRepoArgs carries optional fields for PATCH /api/v1/repos/{owner}/{name}.
+// Pointer fields let the caller omit unset values from the wire payload, so the
+// server only patches what was explicitly requested.
+type EditRepoArgs struct {
+	Archived    *bool   `json:"archived,omitempty"`
+	Description *string `json:"description,omitempty"`
+	Private     *bool   `json:"private,omitempty"`
+	Website     *string `json:"website,omitempty"`
+	Template    *bool   `json:"template,omitempty"`
+}
+
+func (c *Client) EditRepo(ctx context.Context, owner, name string, args EditRepoArgs) (*Repo, error) {
+	body, err := json.Marshal(args)
+	if err != nil {
+		return nil, fmt.Errorf("marshal edit args: %w", err)
+	}
+	path := fmt.Sprintf("/api/v1/repos/%s/%s", owner, name)
+	resp, status, err := c.PatchJSON(ctx, path, body)
+	if err != nil {
+		return nil, err
+	}
+	if err := MapStatus(status, resp); err != nil {
+		return nil, err
+	}
+	var r Repo
+	if err := json.Unmarshal(resp, &r); err != nil {
+		return nil, err
+	}
+	return &r, nil
+}

internal/mcp/server.go

@@ -56,7 +56,6 @@ func (s *Server) handlePOST(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	// initialize is the only method allowed without a session.
 	if req.Method == "initialize" {
 		sid := s.opts.Sessions.Issue()
 		w.Header().Set("Mcp-Session-Id", sid)
@@ -68,11 +67,12 @@ func (s *Server) handlePOST(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 
-	sid := r.Header.Get("Mcp-Session-Id")
+	// Mcp-Session-Id is advisory: we issue one on initialize and accept it back,
-	if !s.opts.Sessions.Valid(sid) {
+	// but every tool the gitea-mcp server exposes is stateless single-shot, so
-		http.Error(w, "missing or invalid Mcp-Session-Id", http.StatusBadRequest)
+	// we do not gate non-initialize calls on it. The claude.ai connector's
-		return
+	// transport proxy is observed to not propagate the session header reliably,
-	}
+	// and the spec allows servers to be sessionless. Compare with brain-mcp /
+	// supervisor-mcp, which never required a session at all.
 
 	switch req.Method {
 	case "tools/list":

internal/mcp/server_test.go

@@ -57,14 +57,22 @@ func TestInitialize(t *testing.T) {
 	assert.Equal(t, "gitea-mcp", si["name"])
 }
 
-func TestPostWithoutSessionRejected(t *testing.T) {
+func TestPostWithoutSessionAccepted(t *testing.T) {
+	// gitea-mcp tools are stateless single-shot; Mcp-Session-Id is advisory.
+	// claude.ai's MCP transport proxy is observed to not propagate the
+	// session header reliably, so non-initialize calls must work without it.
 	srv := newServer(t)
 	rr := postJSON(t, srv, map[string]any{
 		"jsonrpc": "2.0",
 		"id":      2,
 		"method":  "tools/list",
 	}, "")
-	require.Equal(t, http.StatusBadRequest, rr.Code)
+	require.Equal(t, http.StatusOK, rr.Code)
+
+	var resp map[string]any
+	require.NoError(t, json.Unmarshal(rr.Body.Bytes(), &resp))
+	result := resp["result"].(map[string]any)
+	assert.Contains(t, result, "tools")
 }
 
 func TestServerWithOriginAllowlistRejectsBadOrigin(t *testing.T) {

internal/tools/create_project_from_template.go

@@ -45,14 +45,15 @@ func NewCreateProjectFromTemplate(c *gitea.Client, a *allowlist.Allowlist, tmplO
 func (t *CreateProjectFromTemplate) Descriptor() registry.ToolDescriptor {
 	return registry.ToolDescriptor{
 		Name:        "create_project_from_template",
-		Description: "Create a new project repo from the template, applying placeholder substitutions to known files.",
+		Description: "Create a new project repo from a template, applying placeholder substitutions to known files. Defaults to the server-configured template; pass template_name to override (e.g. template-go-agent).",
 		InputSchema: json.RawMessage(`{
 			"type":"object",
 			"properties":{
 				"owner":{"type":"string"},
 				"name":{"type":"string","pattern":"^[a-z][a-z0-9-]{1,38}[a-z0-9]$"},
 				"description":{"type":"string"},
-				"private":{"type":"boolean"}
+				"private":{"type":"boolean"},
+				"template_name":{"type":"string","description":"Template repo name to generate from. Defaults to the server-configured template."}
 			},
 			"required":["owner","name"]
 		}`),
@@ -64,6 +65,7 @@ type createProjectArgs struct {
 	Name         string `json:"name"`
 	Description  string `json:"description"`
 	Private      bool   `json:"private"`
+	TemplateName string `json:"template_name"`
 }
 
 type createProjectResult struct {
@@ -91,13 +93,20 @@ func (t *CreateProjectFromTemplate) Call(ctx context.Context, raw json.RawMessag
 		return nil, fmt.Errorf("name %q does not match pattern %s: %w", args.Name, nameRe.String(), gitea.ErrValidation)
 	}
 
+	// Resolve template: per-call override takes precedence over the
+	// server-configured default. Owner stays server-configured.
+	tmplName := args.TemplateName
+	if tmplName == "" {
+		tmplName = t.templateName
+	}
+
 	// Verify template exists and is marked as a template repo.
-	tmpl, err := t.c.GetRepo(ctx, t.templateOwner, t.templateName)
+	tmpl, err := t.c.GetRepo(ctx, t.templateOwner, tmplName)
 	if err != nil {
 		return nil, fmt.Errorf("template lookup: %w", err)
 	}
 	if !tmpl.Template {
-		return nil, fmt.Errorf("repo %s/%s is not marked as template: %w", t.templateOwner, t.templateName, gitea.ErrValidation)
+		return nil, fmt.Errorf("repo %s/%s is not marked as template: %w", t.templateOwner, tmplName, gitea.ErrValidation)
 	}
 
 	// Verify destination doesn't already exist.
@@ -108,7 +117,7 @@ func (t *CreateProjectFromTemplate) Call(ctx context.Context, raw json.RawMessag
 	}
 
 	// Generate repo from template.
-	newRepo, err := t.c.GenerateFromTemplate(ctx, t.templateOwner, t.templateName, gitea.GenerateFromTemplateArgs{
+	newRepo, err := t.c.GenerateFromTemplate(ctx, t.templateOwner, tmplName, gitea.GenerateFromTemplateArgs{
 		Owner:       args.Owner,
 		Name:        args.Name,
 		Description: args.Description,

internal/tools/create_project_from_template_test.go

@@ -122,6 +122,62 @@ func TestCreateProjectHappyPath(t *testing.T) {
 	assert.Empty(t, out.PartialFailure)
 }
 
+// TestCreateProjectTemplateNameOverride (issue #24): per-call template_name overrides the
+// server-configured default, so the same binary can generate from template-go-web or
+// template-go-agent without restart.
+func TestCreateProjectTemplateNameOverride(t *testing.T) {
+	var templateLookups, generateCalls []string
+
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.Header().Set("Content-Type", "application/json")
+		switch {
+		case r.Method == http.MethodGet && r.URL.Path == "/api/v1/repos/mathias/template-go-agent":
+			templateLookups = append(templateLookups, "template-go-agent")
+			_, _ = w.Write([]byte(newTemplateRepoJSON("template-go-agent", true)))
+
+		case r.Method == http.MethodGet && r.URL.Path == "/api/v1/repos/mathias/template-go-web":
+			templateLookups = append(templateLookups, "template-go-web")
+			_, _ = w.Write([]byte(newTemplateRepoJSON("template-go-web", true)))
+
+		case r.Method == http.MethodGet && r.URL.Path == "/api/v1/repos/mathias/new-agent":
+			w.WriteHeader(http.StatusNotFound)
+			_, _ = w.Write([]byte(`{"message":"not found"}`))
+
+		case r.Method == http.MethodPost && strings.HasSuffix(r.URL.Path, "/generate"):
+			generateCalls = append(generateCalls, r.URL.Path)
+			w.WriteHeader(http.StatusCreated)
+			_, _ = w.Write([]byte(newGeneratedRepoJSON("new-agent")))
+
+		case r.Method == http.MethodGet && strings.HasPrefix(r.URL.Path, "/api/v1/repos/mathias/new-agent/contents/"):
+			filePath := strings.TrimPrefix(r.URL.Path, "/api/v1/repos/mathias/new-agent/contents/")
+			_, _ = w.Write([]byte(fileContentsJSON(filePath)))
+
+		case r.Method == http.MethodPut && strings.HasPrefix(r.URL.Path, "/api/v1/repos/mathias/new-agent/contents/"):
+			filePath := strings.TrimPrefix(r.URL.Path, "/api/v1/repos/mathias/new-agent/contents/")
+			w.WriteHeader(http.StatusOK)
+			_, _ = w.Write([]byte(fileWriteResultJSON(filePath)))
+
+		default:
+			t.Errorf("unexpected request: %s %s", r.Method, r.URL.Path)
+			w.WriteHeader(http.StatusNotFound)
+		}
+	}))
+	defer srv.Close()
+
+	// Server is configured with template-go-web as the default; call overrides to template-go-agent.
+	tool := newCreateProjectTool(srv.URL)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"new-agent","template_name":"template-go-agent"}`,
+	))
+	require.NoError(t, err)
+
+	assert.Equal(t, []string{"template-go-agent"}, templateLookups,
+		"override must direct the template lookup, not the server default")
+	require.Len(t, generateCalls, 1)
+	assert.Equal(t, "/api/v1/repos/mathias/template-go-agent/generate", generateCalls[0],
+		"override must direct the /generate call too")
+}
+
 // TestCreateProjectNameRegexFailure: invalid name returns ErrValidation without hitting network.
 func TestCreateProjectNameRegexFailure(t *testing.T) {
 	tool := tools.NewCreateProjectFromTemplate(

internal/tools/pr_files_diff.go

@@ -143,7 +143,13 @@ func splitUnifiedDiff(d []byte) map[string][]byte {
 
 	flush := func() {
 		if currentFile != "" {
-			m[currentFile] = current.Bytes()
+			// Copy: bytes.Buffer.Bytes() returns the internal slice,
+			// which Reset() then reuses. Without the copy, every map
+			// entry ends up aliased to the last file's data.
+			b := current.Bytes()
+			cp := make([]byte, len(b))
+			copy(cp, b)
+			m[currentFile] = cp
 			current.Reset()
 		}
 	}

internal/tools/pr_files_diff_test.go

@@ -97,6 +97,47 @@ func TestPRFilesDiffSmall(t *testing.T) {
 	assert.ElementsMatch(t, fileNames, paths)
 }
 
+// Regression for issue #25: every file's diff entry must contain its OWN diff,
+// not a shared buffer pointing at the last file. Prior bug: splitUnifiedDiff
+// flushed bytes.Buffer.Bytes() into the map without copying, so every entry
+// aliased the buffer's backing array and showed the last file's content.
+func TestPRFilesDiffPerFileIsolation(t *testing.T) {
+	fileNames := []string{"alpha.go", "beta.go", "gamma.go", "delta.go"}
+	rawDiff := buildDiff(fileNames, 5)
+	filesJSON := buildFilesJSON(fileNames, 5)
+
+	srv := newPRFilesDiffServer(t, filesJSON, rawDiff)
+	defer srv.Close()
+
+	tool := tools.NewPRFilesDiff(gitea.NewClient(srv.URL, "tok"), allowlist.New([]string{"o"}))
+	result, err := tool.Call(context.Background(), json.RawMessage(`{"owner":"o","name":"r","number":1}`))
+	require.NoError(t, err)
+
+	var out struct {
+		Files []struct {
+			Path string `json:"path"`
+			Diff string `json:"diff"`
+		} `json:"files"`
+	}
+	require.NoError(t, json.Unmarshal(result, &out))
+	require.Len(t, out.Files, len(fileNames))
+
+	for _, f := range out.Files {
+		expected := fmt.Sprintf("diff --git a/%s b/%s", f.Path, f.Path)
+		assert.Contains(t, f.Diff, expected,
+			"file %s diff must contain its own header, got: %.80q", f.Path, f.Diff)
+		// No other file's header should leak in.
+		for _, other := range fileNames {
+			if other == f.Path {
+				continue
+			}
+			otherHeader := fmt.Sprintf("diff --git a/%s b/%s", other, other)
+			assert.NotContains(t, f.Diff, otherHeader,
+				"file %s diff must NOT contain %s's header", f.Path, other)
+		}
+	}
+}
+
 func TestPRFilesDiffPerFileTruncated(t *testing.T) {
 	// One file with a 30KB diff (each "+abcdefghij\n" = 12 bytes; 30KB / 12 ≈ 2560 lines).
 	fileNames := []string{"bigfile.go"}

internal/tools/repo_update.go

@@ -0,0 +1,91 @@
+package tools
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/allowlist"
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/gitea"
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/registry"
+)
+
+type RepoUpdate struct {
+	c *gitea.Client
+	a *allowlist.Allowlist
+}
+
+func NewRepoUpdate(c *gitea.Client, a *allowlist.Allowlist) *RepoUpdate {
+	return &RepoUpdate{c: c, a: a}
+}
+
+func (t *RepoUpdate) Descriptor() registry.ToolDescriptor {
+	return registry.ToolDescriptor{
+		Name: "repo_update",
+		Description: "Update repository metadata (description, visibility, default branch, website, archived, template). " +
+			"Only fields explicitly set in the call are patched. " +
+			"WARNING: private=false exposes the repo publicly — verify intent before calling.",
+		InputSchema: json.RawMessage(`{
+			"type":"object",
+			"properties":{
+				"owner":{"type":"string"},
+				"name":{"type":"string"},
+				"description":{"type":"string"},
+				"private":{"type":"boolean","description":"Toggle visibility. false makes the repo public."},
+				"website":{"type":"string","description":"Homepage URL"},
+				"default_branch":{"type":"string","description":"Rename the default branch"},
+				"archived":{"type":"boolean","description":"Mark repo as archived (read-only)."},
+				"template":{"type":"boolean","description":"Toggle template-repo flag"},
+				"confirm":{"type":"string","description":"Required when setting private=false. Must equal the repo name."}
+			},
+			"required":["owner","name"]
+		}`),
+	}
+}
+
+type repoUpdateArgs struct {
+	Owner         string  `json:"owner"`
+	Name          string  `json:"name"`
+	Description   *string `json:"description,omitempty"`
+	Private       *bool   `json:"private,omitempty"`
+	Website       *string `json:"website,omitempty"`
+	DefaultBranch *string `json:"default_branch,omitempty"`
+	Archived      *bool   `json:"archived,omitempty"`
+	Template      *bool   `json:"template,omitempty"`
+	Confirm       string  `json:"confirm"`
+}
+
+func (t *RepoUpdate) Call(ctx context.Context, raw json.RawMessage) (json.RawMessage, error) {
+	var args repoUpdateArgs
+	if err := parseArgs(raw, &args); err != nil {
+		return nil, err
+	}
+	if err := t.a.Check(args.Owner); err != nil {
+		return nil, err
+	}
+
+	// Making a repo public is a significant action — require explicit confirmation.
+	if args.Private != nil && !*args.Private {
+		if args.Confirm != args.Name {
+			return nil, fmt.Errorf("setting private=false makes the repo public: set confirm=%q to proceed", args.Name)
+		}
+	}
+
+	if args.Description == nil && args.Private == nil && args.Website == nil &&
+		args.DefaultBranch == nil && args.Archived == nil && args.Template == nil {
+		return nil, fmt.Errorf("at least one updatable field must be set: %w", gitea.ErrValidation)
+	}
+
+	r, err := t.c.UpdateRepo(ctx, args.Owner, args.Name, gitea.UpdateRepoArgs{
+		Description:   args.Description,
+		Private:       args.Private,
+		Website:       args.Website,
+		DefaultBranch: args.DefaultBranch,
+		Archived:      args.Archived,
+		Template:      args.Template,
+	})
+	if err != nil {
+		return nil, err
+	}
+	return textOK(r)
+}

internal/tools/repo_update_test.go

@@ -0,0 +1,139 @@
+package tools_test
+
+import (
+	"context"
+	"encoding/json"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/allowlist"
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/gitea"
+	"gitea.d-ma.be/mathias/gitea-mcp/internal/tools"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func newRepoUpdateTool(srvURL string) *tools.RepoUpdate {
+	return tools.NewRepoUpdate(gitea.NewClient(srvURL, "tok"), allowlist.New([]string{"mathias"}))
+}
+
+// TestRepoUpdateArchive: happy path — set archived=true.
+func TestRepoUpdateArchive(t *testing.T) {
+	var patchedBody []byte
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		require.Equal(t, http.MethodPatch, r.Method)
+		require.Equal(t, "/api/v1/repos/mathias/old-svc", r.URL.Path)
+		patchedBody, _ = io.ReadAll(r.Body)
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"name":"old-svc","full_name":"mathias/old-svc","default_branch":"main","template":false,"private":false}`))
+	}))
+	defer srv.Close()
+
+	tool := newRepoUpdateTool(srv.URL)
+	result, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"old-svc","archived":true}`,
+	))
+	require.NoError(t, err)
+
+	// Wire payload only contains the field that was actually set.
+	var sent map[string]any
+	require.NoError(t, json.Unmarshal(patchedBody, &sent))
+	assert.Equal(t, true, sent["archived"])
+	assert.NotContains(t, sent, "description")
+	assert.NotContains(t, sent, "private")
+	assert.NotContains(t, sent, "website")
+	assert.NotContains(t, sent, "template")
+
+	var repo gitea.Repo
+	require.NoError(t, json.Unmarshal(result, &repo))
+	assert.Equal(t, "mathias/old-svc", repo.FullName)
+}
+
+// TestRepoUpdateMultipleFields: set description + template flag in one call.
+func TestRepoUpdateMultipleFields(t *testing.T) {
+	var patchedBody []byte
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		patchedBody, _ = io.ReadAll(r.Body)
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"name":"template-go-agent","full_name":"mathias/template-go-agent","description":"Go agent template","template":true}`))
+	}))
+	defer srv.Close()
+
+	tool := newRepoUpdateTool(srv.URL)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"template-go-agent","description":"Go agent template","template":true}`,
+	))
+	require.NoError(t, err)
+
+	var sent map[string]any
+	require.NoError(t, json.Unmarshal(patchedBody, &sent))
+	assert.Equal(t, "Go agent template", sent["description"])
+	assert.Equal(t, true, sent["template"])
+	assert.NotContains(t, sent, "archived")
+	assert.NotContains(t, sent, "private")
+}
+
+// TestRepoUpdateNoFieldsRejected: zero updatable fields → validation error before network.
+func TestRepoUpdateNoFieldsRejected(t *testing.T) {
+	tool := tools.NewRepoUpdate(
+		gitea.NewClient("http://unused", ""),
+		allowlist.New([]string{"mathias"}),
+	)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"some-repo"}`,
+	))
+	require.Error(t, err)
+	assert.ErrorIs(t, err, gitea.ErrValidation)
+}
+
+// TestRepoUpdateMakePublic: explicit private=false is allowed; wire payload carries the false.
+// (The destructive nature is warned about in the tool description, not blocked by the tool.)
+func TestRepoUpdateMakePublic(t *testing.T) {
+	var patchedBody []byte
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		patchedBody, _ = io.ReadAll(r.Body)
+		w.Header().Set("Content-Type", "application/json")
+		_, _ = w.Write([]byte(`{"name":"open-repo","full_name":"mathias/open-repo","private":false}`))
+	}))
+	defer srv.Close()
+
+	tool := newRepoUpdateTool(srv.URL)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"open-repo","private":false}`,
+	))
+	require.NoError(t, err)
+
+	var sent map[string]any
+	require.NoError(t, json.Unmarshal(patchedBody, &sent))
+	assert.Equal(t, false, sent["private"])
+}
+
+// TestRepoUpdateAllowlistRejects: owner outside allowlist denied without network call.
+func TestRepoUpdateAllowlistRejects(t *testing.T) {
+	tool := tools.NewRepoUpdate(
+		gitea.NewClient("http://unused", ""),
+		allowlist.New([]string{"mathias"}),
+	)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"evil","name":"some-repo","archived":true}`,
+	))
+	require.Error(t, err)
+}
+
+// TestRepoUpdateUpstreamError: server 500 propagates as ErrUpstream.
+func TestRepoUpdateUpstreamError(t *testing.T) {
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) {
+		w.WriteHeader(http.StatusInternalServerError)
+		_, _ = w.Write([]byte(`{"message":"internal"}`))
+	}))
+	defer srv.Close()
+
+	tool := newRepoUpdateTool(srv.URL)
+	_, err := tool.Call(context.Background(), json.RawMessage(
+		`{"owner":"mathias","name":"some-repo","archived":true}`,
+	))
+	require.Error(t, err)
+	assert.ErrorIs(t, err, gitea.ErrUpstream)
+}

scripts/context-sync.sh

@@ -0,0 +1,201 @@
+#!/usr/bin/env bash
+# Generates harness-specific context files from .context/PROJECT.md
+# Project-level script — run from a project directory.
+#
+# For Claude Code: generates project-only CLAUDE.md (it inherits root via tree walk)
+# For everything else: concatenates root AGENT.md + project PROJECT.md
+#
+# Usage: ./scripts/context-sync.sh [--force] [adapter...]
+# Task:  task context:sync
+#
+# Override root context: ROOT_CONTEXT=~/dev/.context/AGENT.md ./scripts/context-sync.sh
+
+set -euo pipefail
+
+# Parse --force flag and collect adapter names separately
+FORCE=false
+ADAPTERS=()
+for _arg in "$@"; do
+  case "$_arg" in
+    --force) FORCE=true ;;
+    *) ADAPTERS+=("$_arg") ;;
+  esac
+done
+
+PROJECT_FILE=".context/PROJECT.md"
+
+# Walk up to find root .context/AGENT.md
+find_root_context() {
+  local dir
+  dir="$(pwd)"
+  while [ "$dir" != "/" ]; do
+    dir="$(dirname "$dir")"
+    if [ -f "$dir/.context/AGENT.md" ]; then
+      echo "$dir/.context/AGENT.md"
+      return
+    fi
+  done
+  echo ""
+}
+
+ROOT_CONTEXT="${ROOT_CONTEXT:-$(find_root_context)}"
+
+if [ ! -f "$PROJECT_FILE" ]; then
+  echo "Error: $PROJECT_FILE not found. Are you in a project root?"
+  exit 1
+fi
+
+# Pre-flight: reject unfilled {{...}} placeholders unless --force
+if [ "$FORCE" = false ]; then
+  _placeholders=$(grep -n '{{[^}]*}}' "$PROJECT_FILE" 2>/dev/null || true)
+  if [ -n "$_placeholders" ]; then
+    echo "Error: unfilled placeholders in $PROJECT_FILE:" >&2
+    while IFS= read -r _match; do
+      _lineno="${_match%%:*}"
+      _content="${_match#*:}"
+      _token=$(printf '%s' "$_content" | grep -o '{{[^}]*}}' | head -1)
+      echo "  $PROJECT_FILE:$_lineno: unfilled placeholder $_token" >&2
+    done <<< "$_placeholders"
+    echo "" >&2
+    echo "Fill these placeholders, then re-run: task context:sync" >&2
+    echo "To bypass validation: bash scripts/context-sync.sh --force" >&2
+    exit 1
+  fi
+fi
+
+if [ -n "$ROOT_CONTEXT" ] && [ -f "$ROOT_CONTEXT" ]; then
+  echo "  Root context: $ROOT_CONTEXT"
+else
+  echo "  No root AGENT.md found (project context only)"
+fi
+
+# Emit root context + separator
+root_block() {
+  if [ -n "$ROOT_CONTEXT" ] && [ -f "$ROOT_CONTEXT" ]; then
+    cat "$ROOT_CONTEXT"
+    echo ""
+    echo "---"
+    echo ""
+  fi
+}
+
+# ── Claude Code ──────────────────────────────────────────────
+# Claude Code walks up the tree — it finds ~/dev/CLAUDE.md automatically.
+# Project-level CLAUDE.md only needs project-specific context.
+generate_claude() {
+  cat "$PROJECT_FILE" > CLAUDE.md
+  echo "  → CLAUDE.md (project-only; Claude Code inherits root)"
+}
+
+# ── AGENTS.md (Crush, Pi, Antigravity) ──────────────────────
+# These tools read AGENTS.md from cwd but don't walk up.
+# Concatenate root + project.
+generate_agents() {
+  { root_block; cat "$PROJECT_FILE"; } > AGENTS.md
+  echo "  → AGENTS.md (root + project; Crush, Pi, Antigravity)"
+}
+
+# ── Cursor ───────────────────────────────────────────────────
+generate_cursor() {
+  {
+    echo "# Cursor rules — auto-generated"
+    echo "# Do not edit. Run: task context:sync"
+    echo ""
+    root_block
+    cat "$PROJECT_FILE"
+  } > .cursorrules
+  echo "  → .cursorrules (root + project)"
+}
+
+# ── Aider ────────────────────────────────────────────────────
+generate_aider() {
+  { root_block; cat "$PROJECT_FILE"; } > .aider.conventions.md
+  if [ ! -f .aider.conf.yml ]; then
+    cat > .aider.conf.yml << 'YAML'
+read: .aider.conventions.md
+auto-commits: false
+YAML
+  fi
+  echo "  → .aider.conventions.md (root + project)"
+}
+
+# ── Generic system prompt (Open WebUI, Mods, etc.) ──────────
+generate_system_prompt() {
+  {
+    echo "You are a coding assistant working on a specific project."
+    echo "Follow all conventions from both the root agent context and project context."
+    echo ""
+    echo "---"
+    echo ""
+    root_block
+    cat "$PROJECT_FILE"
+    echo ""
+    echo "---"
+  } > .context/system-prompt.txt
+  echo "  → .context/system-prompt.txt (root + project)"
+}
+
+# ── MCP config ───────────────────────────────────────────────
+generate_mcp() {
+  # Ensure baseline file exists with project-specific knowledge server
+  if [ ! -f .context/mcp.json ]; then
+    cat > .context/mcp.json << 'JSON'
+{
+  "mcpServers": {
+    "knowledge": {
+      "url": "http://localhost:3100/mcp",
+      "description": "Project knowledge base — vector + graph retrieval"
+    }
+  }
+}
+JSON
+  fi
+
+  # Merge root mcp-servers.json if found alongside root AGENT.md
+  local root_mcp=""
+  if [ -n "$ROOT_CONTEXT" ] && [ -f "$ROOT_CONTEXT" ]; then
+    local candidate
+    candidate="$(dirname "$ROOT_CONTEXT")/mcp-servers.json"
+    [ -f "$candidate" ] && root_mcp="$candidate"
+  fi
+
+  if [ -z "$root_mcp" ]; then
+    echo "  → .context/mcp.json (exists, no root mcp-servers.json found)"
+    return
+  fi
+
+  # Root servers take precedence over project entries on key conflict
+  local root_servers count updated
+  root_servers=$(jq '.servers' "$root_mcp")
+  count=$(printf '%s' "$root_servers" | jq 'keys | length')
+  updated=$(jq --argjson root "$root_servers" \
+    '.mcpServers = (.mcpServers + $root)' \
+    .context/mcp.json)
+  printf '%s\n' "$updated" > .context/mcp.json
+  echo "  → .context/mcp.json (merged $count root servers)"
+}
+
+echo "Syncing project context from $PROJECT_FILE..."
+
+if [ ${#ADAPTERS[@]} -eq 0 ]; then
+  generate_claude
+  generate_agents
+  generate_cursor
+  generate_aider
+  generate_system_prompt
+  generate_mcp
+else
+  for adapter in "${ADAPTERS[@]}"; do
+    case "$adapter" in
+      claude)  generate_claude ;;
+      agents)  generate_agents ;;
+      cursor)  generate_cursor ;;
+      aider)   generate_aider ;;
+      prompt|system|openwebui|owui|generic) generate_system_prompt ;;
+      mcp)     generate_mcp ;;
+      *) echo "Unknown adapter: $adapter" >&2; exit 1 ;;
+    esac
+  done
+fi
+
+echo "Done."