agentleFS
sergeeeysergeeey/Claude-cod-top-2026/skills/core/reference-registry/SKILL.md

reference-registry

Skill9 starsChanged 6 months agoMITView on GitHub

Each project can have a references.md file in .claude/memory/ with structured external links. This skill teaches Claude to use them systematically.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use WebFetch with a specific LOOK FOR: prompt from the entry.

    line 47
  2. Distinctive:

    Do NOT fetch entire repos — fetch specific files/sections.

    line 48
  3. Distinctive:

    Prefer official docs over blog posts — blogs drift from reality

    line 106

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- BSV — Brief Skill View | поиск: BSV Скил : reference-registry TL;DR : Реестр решений и паттернов проекта Вызов : 'добавь в registry', 'что в реестре' НЕ для : Глобальные паттерны -->

name: reference-registry description: > [STATUS: review] [CONFIDENCE: high] [REVIEWED: 2026-03-17] Systematic external reference lookup. Reads project references.md, suggests relevant URLs for current task, fetches on demand. Triggers: reference, check external, how others solve, compare with, look up, competitor, alternative implementation, benchmark.

Skill: Reference Registry

Project References (auto-injected)

!`cat .claude/memory/references.md 2>/dev/null | head -50`

When to Load

  • User mentions "check how X does it", "compare with", "look up"
  • Task involves design decisions that benefit from external examples
  • User explicitly says "check references" or "what do our refs say"

How It Works

Each project can have a references.md file in .claude/memory/ with structured external links. This skill teaches Claude to use them systematically.

Workflow

1. Check if references exist

Read .claude/memory/references.md (project-level)
  OR ~/.claude/memory/references.md (global)

2. Match current task to reference entries

Each entry has a WHEN: field — match it against the current task context.

3. Fetch only what's needed

Use WebFetch with a specific LOOK FOR: prompt from the entry. Do NOT fetch entire repos — fetch specific files/sections.

4. Report with evidence markers

[EXTERNAL-REF] From <source>: <finding>
[EXTERNAL-DIFF] Our approach differs: <comparison>

references.md Format

# External References

## Category Name

- [Short Name](https://url/to/specific/file)
  WHEN: <conditions when this reference is relevant>
  LOOK FOR: <what to extract from the URL>
  NOTES: <optional context, caveats, last verified date>

Example

## Architecture Patterns

- [Circuit Breaker in Polly](https://github.com/App-vNext/Polly/blob/main/src/Polly.Core/CircuitBreaker/README.md)
  WHEN: modifying circuit breaker logic, adding new states
  LOOK FOR: state transition rules, half-open behavior, success threshold
  NOTES: .NET implementation but patterns are universal

## Competitor Configs

- [Trail of Bits Claude Config](https://github.com/trailofbits/are-we-right)
  WHEN: comparing security approaches, hook architecture
  LOOK FOR: how they handle prompt injection, what hooks they use
  NOTES: Security-focused team, good baseline for comparison

Rules

  1. Never fetch without purpose — always have a LOOK FOR goal
  2. Cache mentally — if you fetched a URL this session, don't re-fetch
  3. Prefer specific files over repo roots (.../blob/main/src/X.py > .../tree/main)
  4. Mark findings with [EXTERNAL-REF] so user knows the source
  5. Don't assume freshness — URLs may be outdated. Note last verified date if available

Anti-Patterns

Don'tWhyDo Instead
Fetch entire reposWastes tokens, impreciseFetch specific files
Memorize URL contentContent changesRe-fetch when needed
Trust without verificationExternal code may have bugsCross-check with our tests
Fetch on every taskMost tasks don't need refsOnly when WHEN matches

Gotchas

  • References older than 6 months may be outdated — always check version
  • Prefer official docs over blog posts — blogs drift from reality

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.