bb

Files (1)

• SKILL.md

SKILL.md (first lines)

# bb

Read the user's request, match it below, and proceed with the matched skill — read its
SKILL.md and follow it. When a playbook-bearing skill matches, copy the playbook's steps
into your todo list verbatim; a skipped step stays listed with `skip: <reason>`. Nothing
matches: say so and work from the principles directly rather than forcing a fit.

## The map

**Understand something**
- How a repo works, onboarding, architecture → `repo-wiki`
- Explain a diff/PR richly → `explain-diff-html`; dead-simple explanation → `eli5`

**Design and plan**
- Stress-test a plan or idea → `grill-me` (or `grill-with-docs` to grow CONTEXT.md +
  ADRs while at it)
- Conversation is settled, make it a spec → `to-spec-bd` (bd epic)
- Spec into workable tickets with edges → `to-issues-bd`

**Build**
- Build a ticket/feature/refactor → `implement` (feature + refactoring playbooks
  inside; lazy-ladder applies)
- Something's broken → `diagnose`
- Test-first mechanics → `tdd`

**Check**
- Review a diff, want a real verdict → `review`
- Auth/user data/deps in the diff → `sec-review` too
- Prove the app works like a user → the repo's `verify-<app>` skill; create it with
  `verify-setup`, keep it honest with `verify-maintain`
- Watchable proof → `demo`

**Scale up**
- Many tickets in parallel → `orchestrate`
- Optimize a metric in a loop → `autoresearch`
- Did a skill change actually change behavior → `eval`

**Meta**
- New repo → `repo-setup`
- Writing or editing any skill/AGENTS.md → `writing-for-agents` +
  `docs/authoring-standard.md` in the bbstack repo

The behavioural principles (lazy-ladder, prove-it-works, fix-root-causes, ...) are not
invoked directly — the workflow skills above cite them where they apply.

From baranbingol1/bbstack, MIT.

More from baranbingol1/bbstack