huskies/.huskies/AGENT.md

# Huskies project-local agent guidance

## Session Start & Resume Protocol

### PLAN.md — required for every coder session

At the very start of each coder session, before doing any code exploration, check for `PLAN.md` in the worktree root:

**If `PLAN.md` exists (resuming after a watchdog respawn):**
1. Read `PLAN.md` first — it is your primary orientation document.
2. Only after reading it, call `git_log` / `git_diff` to see commits made since the plan was last updated.
3. Reconcile any divergence between the plan and the current git state, then update the plan.

**If `PLAN.md` is absent (first session on this story):**
1. Write `PLAN.md` before any grep, file read, or exploration tool call.
2. Populate it with what you know from the story ACs alone; add specifics as you discover them.

### What PLAN.md must contain

`PLAN.md` is a living document. Update it after each completed AC or natural unit of work — not only at the start.

**Required trigger:** Before every `wip(...)` commit AND the final commit, update PLAN.md's "Current state" section to reflect what's now done, and tick off completed items in "What's left". This is required, not optional — stale "Current state: No code changes yet" while files are being edited is a process failure. Stage the PLAN.md update in the same commit as the code change it describes.

Required sections:

```markdown
# Plan: Story <id>

## ACs → implementation locations
- AC 1: <exact file path>:<line range> — <one-line description of what changes>
- AC 2: <exact file path>:<line range> — …
…

## Decisions
- <Decision made>: <rationale> — rejected alternative: <what was considered and why it lost>
…

## Current state
<What has been done so far. Reference commit hashes or specific functions completed.>

## What's left
- [ ] <specific remaining task with file path and function name>
…
```

### Non-conforming outputs

A PLAN.md that contains only generic steps like "read the code", "write the code", "run the tests", or leaves file paths as `<TBD>` or unspecified is **non-conforming**. Every AC entry must name a real file path and describe the actual change. Every decision entry must name both the chosen approach and at least one rejected alternative with a reason. A stub plan is worse than no plan — rewrite it with specifics.

## Documentation
Docs live in `website/docs/*.html` (static HTML), **not** Markdown files. When a story asks you to document something, edit the relevant `.html` file in `website/docs/`.

## Configuration files
- Agent config: `.huskies/agents.toml` (preferred) or `[[agent]]` blocks in `.huskies/project.toml`
- Project settings: `.huskies/project.toml`
- Bot credentials: `.huskies/bot.toml` (gitignored — never commit)

## Frontend build
The frontend is embedded into the Rust binary via `rust-embed`. Run `npm run build` in `frontend/` before testing frontend changes, or the embedded assets will be stale.

## Quality gates (all enforced by `script/test`)
1. `npm run build` (frontend)
2. `cargo fmt --all --check`
3. `cargo clippy -- -D warnings`
4. `cargo test`
5. `npm test` (frontend Vitest)

Clippy is zero-tolerance: no warnings allowed. Fix every warning before committing.

## File size
Target a maximum of 800 lines per source file as a soft guide. If a file grows beyond 800 lines, decompose it by concern into smaller modules. Split at natural seams: group related types, functions, or handlers together and move each cohesive group to its own file. This keeps files readable and diffs focused.

## Runtime validation
The `validate_agents` function in `server/src/config.rs` rejects unknown runtimes. Supported values: `"claude-code"` and `"gemini"`. Adding a new runtime requires updating that function.