amnafarzy/ravie/current-docs-guard

Claude Code and agent skill by amnafarzy/ravie

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use this before implementing against any version-sensitive or unverified API: package upgrades, SDK and framework APIs (Next.js, Supabase, Vercel, Three.js, MCP, auth, browser APIs), or any external API you would otherwise write from memory unverified this session.

Install

npx vskill@latest install amnafarzy/ravie/current-docs-guard

SKILL.md

The same SKILL.md was published earlier as amnafarzy/ravie/current-docs-guard.

current-docs-guard

Verify current official documentation before changing version-sensitive APIs. Prevents the #1 AI failure mode: confidently writing code from outdated training data.

Purpose

Before any implementation that touches version-sensitive tools (Supabase, Next.js, Vercel, Three.js, MCP, auth systems), check the current official docs. Not blogs, not courses, not memory — the actual docs at the actual URL.

Process

0. Read relevant rule references first

Before long investigations, read rules/context-hygiene.md to keep the docs check scoped and avoid context bloat. Before version-sensitive work, read the relevant rules/*.md file for the domain, if one exists.

1. Identify what's version-sensitive

Before writing any code, list: which external APIs, SDKs, or tools does this change touch?

2. Check the repo's current versions
bash
cat package.json | grep "[library-name]"
# or
cat pnpm-lock.yaml | grep "[library-name]" | head -5
3. Read the actual docs

Go to the official documentation URL. Not a blog post, not a Stack Overflow answer, not your training data. The docs.

Common doc URLs:

  • Supabase: https://supabase.com/docs
  • Next.js: https://nextjs.org/docs
  • Vercel: https://vercel.com/docs
  • Three.js: https://threejs.org/docs
4. Compare docs against current code

Look for:

  • Deprecated methods the codebase still uses
  • New patterns that replace what's currently implemented
  • Breaking changes between installed version and latest
  • Auth patterns that have changed (very common with Supabase)
5. Document what you found
markdown
## Docs check — [library/API]
**Installed version:** [x.y.z]
**Current docs version:** [a.b.c]
**Breaking changes:** [yes/no — list if yes]
**Deprecated patterns found:** [list]
**Recommended approach:** [what the current docs say to do]
**Migration needed:** [yes/no — scope if yes]
**Confidence:** [HIGH — docs verified / MEDIUM — docs partially checked / LOW — unable to verify]
6. Flag uncertainty explicitly

If you can't verify something from docs, say so. "I believe this API works this way based on training data, but I could not verify it from current docs" is vastly better than silently guessing.

Output format

markdown
# Docs guard report — [what was checked]
## Components checked
-[Library]: v[installed] → docs say v[current]
-[API]: [method checked] — [current/deprecated/changed]
## Stale assumptions avoided
-[Pattern that would have been wrong without checking]
## Recommendation
[What to implement based on current docs]
## Risks
-[Migration risk if upgrading]
-[Compatibility concern]
## Confidence: [HIGH/MEDIUM/LOW]

Hard rules

  • Never rely on training data for version-sensitive API behavior — always verify
  • Never silently upgrade packages without checking migration guides
  • Never use deprecated patterns when docs show the current replacement
  • Never bury uncertainty — flag it explicitly with confidence level
  • Prefer official docs over courses, blogs, summaries, or Stack Overflow

Connects to

  • issue-to-pr — invoked during implementation for version-sensitive work
  • supabase-guardian — complementary: docs-guard checks the API, supabase-guardian checks the schema/RLS
  • decision-log-adr — log decisions to upgrade or stay on current version

Common failure modes

Confidently wrong — AI generates Supabase auth code from 6 months ago. The API changed. Code compiles but auth silently breaks in edge cases. Always verify.

Checking blogs instead of docs — A Medium post from 2024 is not current documentation. Go to the official URL.

Skipping the version check — Docs for v3 look different from v2. If you're on v2, read v2 docs, not latest.

"I'm pretty sure this is right" — If you're "pretty sure" about an API method without checking, flag it as MEDIUM confidence. Let the human decide whether to verify.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
3
7-day trend score
7

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/amnafarzy/ravie/current-docs-guard/badge)](https://verified-skill.com/skills/amnafarzy/ravie/current-docs-guard)

Something wrong with this skill or its page? Tell us.