Adonis0123/adonis-skills / skills/agent-practice-audit
agent-practice-audit
FlaggedScore how someone uses coding agents (Claude Code, Codex, Grok and WorkBuddy session logs, instruction files, hooks, permissions, Codex config) against practices quoted from official vendor docs, and write a local HTML report with coverage, trend and suggestions; also refresh or verify that practice list from official docs with multi-model consensus. Use when the user asks how far their agent usage is from official practice, wants an agent usage check-up or grade, or wants the practice list refreshed, its links checked, or either run on a schedule. Chinese triggers: Agent 用法体检、对照官方最佳实践、我的 agent 用得对不对、给 agent 用法打分、更新最佳实践清单. Not for whether a tool, MCP or skill is worth keeping (dev-environment-roi-audit), for writing or reviewing one CLAUDE.md or AGENTS.md, for general best-practice questions, or for Cursor or Gemini CLI logs, which it does not read. Never edits the user's config.
references/sources.md
View on GitHub# Sources
Refresh mode searches only these official domains. A page outside the list is never added, however good it looks. Third-party blogs, newsletters, X/Twitter, YouTube, Reddit, Medium and course sites are excluded, including posts that quote official pages: follow the quote back to the official page and cite that.
`scripts/check_practices.py` enforces the same list (`ALLOWED_DOMAINS`, `GITHUB_ORGS`). Change both together.
## Allowlist
| Vendor | Domains | Notes |
| --------- | -------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- |
| Anthropic | `code.claude.com`, `docs.anthropic.com`, `docs.claude.com`, `claude.com`, `anthropic.com` (incl. `www.`) | Claude Code docs and `anthropic.com/engineering` posts. |
| OpenAI | `learn.chatgpt.com`, `developers.openai.com`, `openai.com`, `platform.openai.com` | Codex guides, docs and announcements. |
| Google | `geminicli.com`, `github.com/google-gemini/*`, `developers.googleblog.com`, `ai.google.dev` | Gemini CLI docs live on the site and in the `google-gemini/gemini-cli` repository `docs/`. Other GitHub orgs are not allowed. |
| Cursor | `cursor.com`, `docs.cursor.com` | Docs and the official blog. |
| xAI | `docs.x.ai` | Grok docs. |
| Tencent | `codebuddy.ai` (incl. `www.`), `cloud.tencent.com` | CodeBuddy product docs only; other `cloud.tencent.com` products are out of scope. |
## Seed URLs
Start here, then follow in-site links and the vendor's own search. Seeds marked _verified_ back at least one practice in `practices.json` (checked live on the date in `changelog.md`). Seeds marked _seed_ are starting points that refresh has not yet confirmed; fetch them before relying on them.
| Vendor | URL | State |
| --------- | -------------------------------------------------------------- | --------------------------------- |
| Anthropic | https://code.claude.com/docs/en/best-practices | verified |
| Anthropic | https://code.claude.com/docs/en/memory | verified |
| Anthropic | https://code.claude.com/docs/en/costs | verified |
| Anthropic | https://code.claude.com/docs/en/common-workflows | verified |
| Anthropic | https://code.claude.com/docs/en/hooks-guide | seed |
| Anthropic | https://code.claude.com/docs/en/settings | seed |
| Anthropic | https://code.claude.com/docs/en/sub-agents | seed |
| Anthropic | https://www.anthropic.com/engineering | seed (index of engineering posts) |
| OpenAI | https://learn.chatgpt.com/guides/best-practices | verified |
| OpenAI | https://learn.chatgpt.com/docs/sandboxing.md | verified |
| OpenAI | https://learn.chatgpt.com/docs/non-interactive-mode | verified |
| OpenAI | https://learn.chatgpt.com/docs/agent-configuration/agents-md | verified |
| OpenAI | https://developers.openai.com/codex | seed |
| Google | https://geminicli.com/docs/ | seed |
| Google | https://github.com/google-gemini/gemini-cli/tree/main/docs | seed |
| Google | https://developers.googleblog.com/ (search "Gemini CLI") | seed |
| Cursor | https://cursor.com/blog/agent-best-practices | verified |
| Cursor | https://cursor.com/docs | seed |
| xAI | https://docs.x.ai/ | seed |
| Tencent | https://www.codebuddy.ai/docs | seed |
| Tencent | https://cloud.tencent.com/document/product (CodeBuddy section) | seed |
## Offline digests
Every page that a practice cites has `references/sources/<slug>.md`, and each source entry in `practices.json` names it in `source`. The digest holds frontmatter (`slug`, `title`, `publisher`, `url`, `fetchedAt`, `lastVerifiedAt`, `linkStatus` = ok | redirected | unreachable, `finalUrl` when redirected), a 5-12 bullet summary in our own words (no copied paragraphs), and the short verbatim quotes the practices cite. When a page cannot be opened (403, region block, outage), the report keeps the link but marks it unavailable and shows the digest, and verifiers judge against the digest (`basis: digest`). `check_practices.py --fetch --update-digests` refreshes the link state; it never deletes a digest or a practice.
## What counts as a practice
- An instruction a user can follow or check: "keep CLAUDE.md under 200 lines", not "Claude is great at refactoring".
- Stated by the vendor in its own voice. Release notes count when they recommend a behaviour; feature lists alone do not.
- Observable in local logs or config (or honestly `na` with a reason in `detect`).
- Not a duplicate: when two vendors say the same thing, add the second quote as another source on the existing practice.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |