agentleFS
PaulRBergPaulRBerg/dot-agents/skills/cli-coingecko/SKILL.md

cli-coingecko

Skill6 starsChanged 6 days agoMITView on GitHub

This skill is coordination-exempt: skip the ai-coord gate for its declared work. Use the installed CLI's machine-readable command catalog as the authority for supported market-data operations. Do not run interactive cg auth or write config without the user's approval. 2. On an explicitly requested CLI update, run cg update --dry-run. Then run cg update within that authorization. Ask only if the preview reveals an action beyond the requested update. If automatic detection is wrong, pass --method with homebrew, npm, go,…

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use the installed CLI's machine-readable command catalog as the authority for supported market-data operations.

    line 14
  2. Very specific:

    Verify cg exists. Inspect non-interactive auth/tier state:

    line 18
  3. Very specific:

    Do not run interactive cg auth or write config without the user's approval.

    line 26
  4. Distinctive:

    If one prevents completion and a visual result can satisfy the request, immediately read references/open-historical-page.md.

    line 48
  5. Distinctive:

    Do not use the browser fallback for authentication, tier, rate-limit, invalid-input, or ambiguity errors.

    line 61
  6. Distinctive:

    Never expose API keys or send private wallet/account data to market-data endpoints.

    line 66
  7. Very specific:

    When returning control, use ### ⚠️ Rate limited — retry after <time>.

    line 80

≈ “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

CoinGecko CLI

This skill is coordination-exempt: skip the ai-coord gate for its declared work.

Use the installed CLI's machine-readable command catalog as the authority for supported market-data operations.

Workflow

  1. Verify cg exists. Inspect non-interactive auth/tier state:

    command -v cg
    cg status -o json
    cg commands -o json

    Do not run interactive cg auth or write config without the user's approval.

  2. On an explicitly requested CLI update, run cg update --dry-run. Then run cg update within that authorization. Ask only if the preview reveals an action beyond the requested update. If automatic detection is wrong, pass --method with homebrew, npm, go, or script. Recheck cg version and cg commands -o json. Then, when maintaining this skill, record the verified version in references/version.txt.
  3. Select the command, flags, enum values, output formats, endpoint, auth requirement, and paid_only status from cg commands -o json. Use cg <command> --help only when the catalog lacks a needed detail.
  4. Resolve CoinGecko IDs with cg search <term> -o json when the user supplied a name or ambiguous symbol. Do not silently treat symbols as unique.
  5. Preview unfamiliar or quota-sensitive requests with --dry-run. Execute parseable queries with -o json. Use --export only when the user requested a CSV artifact.
  6. For a historical-data request, these CLI failures can trigger the browser fallback:

    • cg is unavailable.
    • cg history fails due to a CLI defect.
    • Its response is demonstrably malformed.

    If one prevents completion and a visual result can satisfy the request, immediately read references/open-historical-page.md. Under those conditions, use its browser fallback.

  7. Present the requested result in the user's format. For ordinary human-readable output, use a compact table. Preserve enough precision for the asset's magnitude.

Boundaries and Defaults

  • Batch IDs in one request when the command supports it. On 429, respect the reported reset/backoff instead of retrying aggressively.
  • Detect paid-only commands before execution. If the current tier cannot serve the request, report that limitation. Under that condition, offer a supported route.
  • Do not use the browser fallback for authentication, tier, rate-limit, invalid-input, or ambiguity errors. It does not satisfy requests for JSON, CSV, or other machine-readable evidence.
  • cg does not cover every CoinGecko endpoint. For unsupported contract-address prices, global stats, NFT detail, GeckoTerminal, or logo metadata, fetch the relevant current API documentation from https://docs.coingecko.com/llms.txt. For those requests, state that the CLI route is unavailable.
  • Never expose API keys or send private wallet/account data to market-data endpoints.

Completion requires the resolved coin/command, successful JSON or requested export evidence, and explicit handling of tier, ambiguity, or rate-limit constraints. A historical-page fallback instead requires a validated URL opened in Chromium plus disclosure of the CLI failure that triggered it.

User-Facing Output

For ordinary human output, lead with ### 🪙 <coin> (<id>). Show only the requested metrics in a compact table. Add one source line with the endpoint plus timestamp, tier, or window when material.

For exports, use ### ✅ Exported <row count> rows. Link the artifact. State the query/window without reproducing the CSV.

When returning control, use ### ⚠️ Rate limited — retry after <time>. Only while actively waiting, use ### ⏳ Rate limited — retrying after <time>. For a paid endpoint, use ### ⚠️ Paid endpoint — current tier: <tier> with one supported route. Return requested JSON/CSV, exact IDs, URLs, values, commands, and diagnostics undecorated.

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.