mirror of https://github.com/garrytan/gstack.git
docs: add Health Stack section to pin /health to the fast path
gstack has no tsconfig/biome/knip/shellcheck, and the full `bun test` script spawns browser integration tests that hang in non-interactive agent shells. Document the fast deterministic tier plus slop-scan so /health completes and targets the real quality gates. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BV2RsGpwAdbDid7yAHBtXK Co-authored-by: Orca <help@stably.ai>
This commit is contained in:
parent
a3259400a3
commit
244fbb8254
15
CLAUDE.md
15
CLAUDE.md
|
|
@ -81,6 +81,21 @@ bun run test:evals # run before shipping — paid, diff-based (~$4/run max)
|
|||
integration tests. `bun run test:evals` runs LLM-judge quality evals and E2E
|
||||
tests via `claude -p`. Both must pass before creating a PR.
|
||||
|
||||
## Health Stack
|
||||
|
||||
The `/health` skill auto-detects nothing useful here: there's no `tsconfig.json`
|
||||
(Bun runs TypeScript directly), no biome/eslint, no knip, and shellcheck usually
|
||||
isn't installed. The real quality gates are the test suite and slop-scan. The full
|
||||
`bun test` script also spawns browser integration tests that hang in non-interactive
|
||||
agent shells, so `/health` targets the fast deterministic tier instead. slop-scan is
|
||||
this repo's linter (see the Slop-scan section below).
|
||||
|
||||
- test: bun test test/skill-validation.test.ts test/gen-skill-docs.test.ts test/hermetic-wiring.test.ts
|
||||
- lint: npx slop-scan scan .
|
||||
|
||||
Run the full `bun test` yourself in a real terminal before a PR — the `/health` entry
|
||||
above is the subset that completes without a browser, not a replacement for it.
|
||||
|
||||
## Project structure
|
||||
|
||||
```
|
||||
|
|
|
|||
Loading…
Reference in New Issue