/gate

Manual gate runner — detects the stack(s) present in a ticket's worktree and runs the corresponding lint/SAST/test gates. Writes structured findings to .tickets/XXXX-<slug>/gate-findings.md.

Ticket resolution

A ticket number argument is required. If none is provided, scan .tickets/ for tickets with status: implementing or status: review-ready. If exactly one exists, use it. Otherwise list candidates and stop.

Stack detection

Inspect the worktree root for these markers and run gates for every stack that matches. A worktree may have multiple stacks.

Marker(s) present	Stack
pyproject.toml, setup.py, setup.cfg, *.py	Python
package.json + tsconfig.json (or *.ts)	TypeScript
package.json (without TS markers)	JavaScript
go.mod	Go
Cargo.toml	Rust

Gates per stack

For each detected stack, run the gate set below. Skip any individual gate whose tool is absent and note "not installed" in findings. Each stack's tools are checked against the changed files (git diff --name-only main) for per-file gates and the full worktree for test/build gates.

Python

Gate	Command (per file or per project)
ruff	ruff check --output-format=concise <files>
mypy	mypy --no-error-summary <files>
bandit	bandit -ll -q -f txt <files>
pytest	pytest -q --no-header

JavaScript / TypeScript

Gate	Command
eslint	npx --no-install eslint --no-color .
tsc	npx --no-install tsc --noEmit (TS only)
tests	npm test --silent

Go

Gate	Command
gofmt	gofmt -l . (any output ⇒ unformatted)
vet	go vet ./...
tests	go test ./...

Rust

Gate	Command
fmt	cargo fmt --check
clippy	cargo clippy -- -D warnings
tests	cargo test --quiet

Output

Write .tickets/XXXX-<slug>/gate-findings.md:

# Gate Findings — XXXX-<slug>

**Run at**: YYYY-MM-DD HH:MM
**Worktree**: .worktrees/XXXX-<slug>
**Stacks detected**: <comma-separated, e.g. "python, typescript">

## <stack>

### <gate>
<concise output or "clean">

Then print a one-line summary: gate: <stack>=<PASS|FAIL: gates-failing> for each stack.

Failure handling

This command does not fix findings — it only records them. The caller (/implement or the lead) decides what to do.