xonecas.io
64875
Mc

mini-coder / commit 4887dc3

refactor: improve review, audit, and release skills per agentskills.io spec

- review: expanded description with trigger keywords, structured body with
  steps, review criteria, and guidelines sections
- audit: restructured into 3 clear phases (code review, multi-provider UI
  probe, report), added compatibility field, better model selection guidance
- release: restructured into numbered steps with clear guardrails, edge case
  handling, and post-push verification flow
- all skills use proper frontmatter with quoted descriptions containing
  trigger keywords for better discoverability

xonecas2026-03-20 4887dc3131ed5a4ad965b3279340872dcbc0c2a2 parent 51586e9

5 files changed

.agents/skills/audit/SKILL.md+15−13
@@ -1,22 +1,24 @@
11 ---
22 name: audit
3-description: Full codebase audit — code review, multi-provider UI testing, and combined report.
3+description: "Full codebase audit — code review, multi-provider UI testing, and combined report. Use when the user asks to audit, test across providers, or validate the full implementation."
4+compatibility: "Requires tmux, bun, git, and active provider credentials (Anthropic OAuth and/or Opencode zen)"
45 ---
56
67 Run a complete audit of the mini-coder implementation.
78
8-## Phase 1 — Code & Architecture Review
9+## Phase 1 — Code & architecture review
910
10-Audit the codebase against the core idea:
11+Audit the codebase against the project's core idea (the source of truth for design):
1112
12-- Question design choices and rethink as needed while staying compliant with the original idea.
13-- Employ KISS, DRY, and YAGNI — no surprise additions or eager optimizations.
14-- Check TODO.md for known issues already tracked.
15-- Verify all documentation matches the idea file.
13+1. Read the core idea section in AGENTS.md (or the system prompt context).
14+2. Walk through each module and question design choices — stay compliant with the original idea.
15+3. Employ KISS, DRY, and YAGNI — flag surprise additions or eager optimizations.
16+4. Check `TODO.md` for known issues already tracked — don't duplicate them.
17+5. Verify documentation matches the idea file.
1618
17-Use `bun run dev` and `tmux` via shell to run one-shot and interactive sessions manually. Analyze console output and UI from a user's perspective.
19+Use `bun run dev` and `tmux` via shell to run one-shot and interactive sessions manually. Analyse console output and UI from a user's perspective.
1820
19-## Phase 2 — Multi-Provider UI Probe
21+## Phase 2 — Multi-provider UI probe
2022
2123 Run `mc` sessions across all SDK paths to capture what real users see.
2224 Do NOT delegate analysis to the models under test.
@@ -29,8 +31,8 @@ Pick cheap options. Verify available models at https://opencode.ai/docs/zen/, th
2931
3032 - `@ai-sdk/anthropic` path → a `zen/claude-*` haiku-class model
3133 - `@ai-sdk/google` path → a `zen/gemini-*` flash-class model
32-- `@ai-sdk/openai` path (responses endpoint) → a `zen/gpt-*` small low cost model
33-- `@ai-sdk/openai-compatible` path → `zen/*` or similar small low cost model
34+- `@ai-sdk/openai` path (responses endpoint) → a `zen/gpt-*` small low-cost model
35+- `@ai-sdk/openai-compatible` path → `zen/*` or similar small low-cost model
3436
3537 Avoid free models due to rate limit issues.
3638 Also use an Anthropic OAuth model for completeness.
@@ -44,9 +46,9 @@ Use prompts that exercise realistic multi-step, multi-turn patterns. Check the d
4446 Write `AUDIT-REPORT.md` in the repo root with these sections:
4547
4648 - **Summary** — how you tested and what you found.
47-- **Code & Architecture** — correctness issues, KISS, DRY, and YAGNI violations, idea misalignment.
49+- **Code & Architecture** — correctness issues, KISS/DRY/YAGNI violations, idea misalignment.
4850 - **UI/UX Alignment** — comparison of actual output vs the core idea's expectations across providers.
4951 - **Recommendations** — split into immediate bugs, code changes, and polish items.
5052
51-Don't just recite the todo items back to the user, report on the audit itself.
53+Don't just recite the todo or known issues items back to the user — report on the audit itself and new findings.
5254 Be concise. The report is analysis only — review findings with the user before making any changes.
.agents/skills/release/SKILL.md+29−23
@@ -1,32 +1,38 @@
11 ---
22 name: release
3-description: Create a new tagged npm release (patch/minor/major)
3+description: "Create a new tagged npm release (patch/minor/major). Use when the user asks to release, publish, bump version, or cut a new version."
4+compatibility: "Requires git, bun, and npm with publish credentials"
45 ---
56
67 Create a new npm release for this repo. The bump type is: $1 (default to "patch" if not provided).
78
8-Steps to follow exactly:
9+## Steps
910
10-1. Normalize bump type to one of `patch|minor|major`; if empty, use `patch`.
11-2. Verify git state before doing anything:
12- - current branch must be `main`
13- - working tree must be clean
14- - run `git fetch origin --tags` and ensure `HEAD` equals `origin/main`
15- If any check fails, stop and report.
16-3. Run the full repo check suit — stop and report failures.
17-4. Read `package.json` and capture both `name` and current `version`.
18-5. Compute the next semver version by bumping the selected part.
19-6. Guardrails before mutating files:
20- - ensure git tag `v<new-version>` does not exist locally or on origin
21- - ensure npm version does not already exist: `npm view <package-name>@<new-version> version`
22- If either exists, stop and report.
23-7. Update the `version` field in `package.json` to `<new-version>`.
24-8. Run `bun run build`.
25-9. Commit all staged and unstaged changes with message: `chore: release v<new-version>`.
26-10. Create annotated tag: `git tag -a v<new-version> -m "<new-version>"`.
27-11. Push explicitly to main and the exact tag:
11+1. **Normalize** the bump type to one of `patch | minor | major`. Default to `patch` if empty.
12+2. **Verify git state** before mutating anything:
13+ - Current branch must be `main`.
14+ - Working tree must be clean (`git status --porcelain` is empty).
15+ - Run `git fetch origin --tags` and ensure `HEAD` equals `origin/main`.
16+ - If any check fails, stop and report.
17+3. **Run checks** — execute the full repo check suite. Stop and report on failures.
18+4. **Read `package.json`** and capture `name` and current `version`.
19+5. **Compute next version** by bumping the selected semver part.
20+6. **Guardrails** before mutating files:
21+ - Ensure git tag `v<new-version>` does not exist locally or on origin.
22+ - Ensure npm version does not already exist: `npm view <name>@<new-version> version`.
23+ - If either exists, stop and report.
24+7. **Update `package.json`** — set the `version` field to `<new-version>`.
25+8. **Build** — run `bun run build`.
26+9. **Commit** all staged and unstaged changes: `chore: release v<new-version>`.
27+10. **Tag** — create annotated tag: `git tag -a v<new-version> -m "<small changelog>"`.
28+11. **Push** explicitly to main and the exact tag:
2829 - `git push origin main`
2930 - `git push origin v<new-version>`
30-12. Post-push verification: run `npm view <package-name>@<new-version> version`.
31- - If it already exists, report that publish already happened and do **not** ask user to publish again.
32- - If it does not exist, ask user to run `npm publish` (passkey auth step).
31+12. **Post-push verification** — run `npm view <name>@<new-version> version`.
32+ - If it already exists, report that publish already happened — do **not** ask to publish again.
33+ - If it does not exist, run `npm publish`.
34+
35+## Edge cases
36+
37+- If the build fails, revert the version bump in `package.json` and stop.
38+- If the push fails, do not retry — report the failure and let the user resolve.
AGENTS.md+1−1
@@ -43,7 +43,7 @@ conventions and not introduce more specs.
4343 - We want to do our best to support and autodiscover community config standards:
4444 - AGENTS.md -> https://agents.md/
4545 - CLAUDE.md (same as AGENTS.md)
46- - Skills -> https://agentskills.io/client-implementation/adding-skills-support (and https://agentskills.io/client-implementation/adding-skills-support.md)
46+ - Skills -> https://agentskills.io/client-implementation/adding-skills-support (and https://agentskills.io/client-implementation/adding-skills-support.md, https://agentskills.io/llms.txt)
4747 - Claude skills too. They claim to follow the spec, we should only need to discover them too.
4848
4949 - Auto discovery of providers via ENV (Example: OPENCODE_API_KEY), or local servers (Example: ollama)
docs/KNOWN_ISSUES.md+0−0
src/cli/bootstrap.ts+21−9
@@ -6,19 +6,31 @@ import { writeln } from "./output.ts";
66
77 const REVIEW_SKILL_CONTENT = `---
88 name: review
9-description: Review recent changes for correctness, code quality, and performance
9+description: "Review recent changes for correctness, code quality, and performance. Use when the user asks to review, check, or audit recent code changes, diffs, or pull requests."
1010 context: fork
1111 ---
12-You are a code reviewer. Review recent changes and provide actionable feedback.
1312
14-Perform a sensible code review:
15-- Correctness: Are the changes in alignment with the goal?
16-- Code quality: Is there duplicate, dead, or bad code patterns introduced?
17-- Is the code performant?
18-- Never flag style choices as bugs, don't be a zealot.
19-- Never flag false positives, check before raising an issue.
13+Review recent changes and provide actionable feedback.
2014
21-Output a small summary with only the issues found. If nothing is wrong, say so.
15+## Steps
16+
17+1. Identify the changes to review — check \`git diff\`, \`git log\`, and staged files.
18+2. Read the changed files and understand the intent behind each change.
19+3. Evaluate each change against the criteria below.
20+4. Output a concise summary with only the issues found. If nothing is wrong, say so.
21+
22+## Review criteria
23+
24+- **Correctness** — Are the changes aligned with their stated goal? Do they introduce bugs or regressions?
25+- **Code quality** — Is there duplicate, dead, or overly complex code? Are abstractions appropriate?
26+- **Performance** — Are there unnecessary allocations, redundant I/O, or algorithmic concerns?
27+- **Edge cases** — Are boundary conditions and error paths handled?
28+
29+## Guidelines
30+
31+- Never flag style choices as bugs — don't be a zealot.
32+- Never flag false positives — verify before raising an issue.
33+- Keep feedback actionable: say what's wrong and suggest a fix.
2234 `;
2335
2436 export function bootstrapGlobalDefaults(): void {