go

Files (62)

• SKILL.md
• advise.md · chunks/
• benchmark.md · chunks/
• critique.md · chunks/
• decompose.checkpoint.md · chunks/
• decompose.extract-one.md · chunks/
• decompose.iterate.md · chunks/
• decompose.md · chunks/
• decompose.wire-app.md · chunks/
• design.kit-boundary.md · chunks/
• design.new.md · chunks/
• document.md · chunks/
• fix.md · chunks/
• implement.md · chunks/
• investigate.debug.md · chunks/
• investigate.explain.md · chunks/
• investigate.extract.md · chunks/
• investigate.migrate.md · chunks/
• investigate.optimize.md · chunks/
• investigate.refactor.md · chunks/
• investigate.review.md · chunks/
• migrate.iterate.md · chunks/
• migrate.plan.md · chunks/
• migrate.unit.md · chunks/
• orient.md · chunks/
• refactor.iterate.md · chunks/
• refactor.plan.md · chunks/
• refactor.propose.md · chunks/
• refactor.unit.md · chunks/
• simplify.md · chunks/
• spec-review.md · chunks/
• triage.md · chunks/
• verify.md · chunks/
• README.md · reference/
• cli.md · reference/
• debugging.md · reference/
• delve.md · reference/
• files.md · reference/
• http.md · reference/
• logging.md · reference/
• modern-go.md · reference/
• packages.md · reference/
• release.md · reference/
• wails.md · reference/
• README.md · rubric/
• README.md · rubric/api-design/
• interfaces.md · rubric/api-design/
• receivers.md · rubric/api-design/
• signatures.md · rubric/api-design/
• surface.md · rubric/api-design/
• checklist.md · rubric/concurrency/
• decisions.md · rubric/
• checklist.md · rubric/documentation/
• checklist.md · rubric/errors/
• control-flow.md · rubric/idioms/
• data-structures.md · rubric/idioms/
• README.md · rubric/naming/
• checklist.md · rubric/naming/
• gauntlet.md · rubric/naming/
• principles.md · rubric/
• …and 2 more

SKILL.md (first lines)

# Go

This skill is a **digraph of small files**, not a rulebook to read top-to-bottom. It merges the
routed workflow, the design rubric, and domain knowledge distilled from
[spf13/go-skills](https://github.com/spf13/go-skills) (MIT).

**The tree:**
- `chunks/` — the workflow: small process steps wired by `## Next` edges.
- `rubric/` — the judgment: what good Go looks like; the gates (`simplify`, `critique`) consult it.
- `reference/` — the knowledge: current-Go currency (`modern-go.md`) and domain leaves (HTTP, CLI,
  Wails, files, releases, logging, packages, debugging), loaded only when the task touches them.

**How to use it:**
1. Start at `chunks/triage.md`. Always.
2. Load **only the one file you need next** — never read the whole skill into context.
3. Each chunk ends with a `## Next` section listing `condition → file` edges. Do the chunk's work,
   then pick the edge that matches what actually happened (your decision, or the user's input) and
   load that file.
4. Triage also flags **domain leaves** (`reference/…`). Carry them forward; the chunk that writes
   or judges code loads them there, not at triage time.
5. The path is **not** decided upfront. It emerges, one edge at a time.

Do not skim all the files. Bulk dilutes focus; one small file applied *now* is what makes the
guidance bind.

**Portability — any model, any harness.** The skill assumes nothing beyond the ability to read
files and run shell commands. If your harness lacks a feature, use the fallback; never skip the step:
- **No sub-agents** → do the work single-threaded: `decompose` uses `chunks/decompose.checkpoint.md`;
  for refactor/migrate sweeps, run each unit yourself and re-orient from the ledger between units;
  for `simplify`'s high-stakes verifier, re-read your revised diff cold in a separate adversarial
  pass before finishing.
- **No hooks** → nothing changes: `chunks/verify.md` runs the full baseline manually; the hook is
  only a backstop.
- **Small context window** → hold at most SKILL.md + the current chunk + one rubric/reference leaf
  at a time. Ledger files (`.go-refactor/`, `.go-decompose/`, `.go-migrate/`) are the memory —
  re-read them, never trust the transcript.

→ Load `chunks/triage.md`.

From vinayprograms/agent-skills, Apache-2.0.

More from vinayprograms/agent-skills