← Browse

@dvcrn/greptile

B

Query, search, and manage repositories indexed by Greptile (AI codebase intelligence). Use when asking questions about a codebase, searching for code patterns, indexing repos for Greptile review, or checking Greptile index status. Requires GREPTILE_TOKEN and a GitHub/GitLab token.

skillclaude

Install

agr install @dvcrn/greptile --target claude

Writes 1 file into .claude/skills/, pinned to git-9d1b3687.

  • .claude/skills/greptile/SKILL.md

Document


name: greptile description: "Query, search, and manage repositories indexed by Greptile (AI codebase intelligence). Use when asking questions about a codebase, searching for code patterns, indexing repos for Greptile review, or checking Greptile index status. Requires GREPTILE_TOKEN and a GitHub/GitLab token."

Greptile Skill

Query and manage Greptile-indexed repositories via the REST API.

Setup

Required environment variables:

  • GREPTILE_TOKEN — Greptile API token (from https://app.greptile.com)
  • GITHUB_TOKEN — GitHub PAT with repo access (alternatively set GREPTILE_GITHUB_TOKEN, or authenticate via gh auth login — the script falls back to gh auth token)

Commands

All commands use scripts/greptile.sh (resolve path relative to this skill directory).

Index a repository

scripts/greptile.sh index owner/repo [branch] [--remote github|gitlab] [--no-reload] [--no-notify]

Default branch: main. Use --no-reload to skip re-indexing if already processed.

Check index status

scripts/greptile.sh status owner/repo [branch] [--remote github|gitlab]

Returns: status (completed/processing/failed), filesProcessed, numFiles.

Query a codebase (AI answer + sources)

scripts/greptile.sh query owner/repo [branch] "How does auth work?" [--genius] [--remote github|gitlab]
  • --genius — slower but smarter analysis (good for PR reviews, architecture questions)
  • Returns: AI-generated answer + source file references with line numbers

Search a codebase (sources only, no AI answer)

scripts/greptile.sh search owner/repo [branch] "payment processing" [--remote github|gitlab]

Returns: ranked list of relevant files, functions, and snippets with line numbers.

Tips

  • Always index a repo before querying/searching it. Check status to confirm indexing is complete.
  • Use query --genius for complex questions (architecture, PR review context, cross-file analysis).
  • Use search when you just need file locations without an AI explanation.
  • For GitLab repos, pass --remote gitlab.
  • Pipe output through jq for formatting: scripts/greptile.sh query ... | jq .
  • Multi-repo queries: not supported by the wrapper; use the API directly with multiple repositories in the body.

Repository README

Describes dvcrn/openclaw-skills-marketplace as a whole, which may contain artifacts other than this one. Where this artifact had no useful description of its own, its summary was taken from here.

OpenClaw Skills Marketplace For Claude

This repository is an automatic conversion of openclaw/skills into the Claude Skills marketplace format so the skills can be installed and used directly in Claude.

It takes the OpenClaw skill corpus and rewrites it into a Claude-compatible marketplace with:

  • a root marketplace manifest at .claude-plugin/marketplace.json
  • one Claude plugin per OpenClaw skill under plugins/
  • a generation report at reports/generate-marketplace.json

Use In Claude

Add this marketplace in Claude with:

/plugin marketplace add dvcrn/openclaw-skills-marketplace

Then install any generated plugin from the marketplace:

/plugin install <plugin-name>@openclaw-skills

Example:

/plugin install 0x-professor--agentic-mcp-server-builder@openclaw-skills

What This Repo Does

The generator reads canonical OpenClaw source skills from:

openclaw-skills/skills/<owner>/<slug>/

Each source skill is expected to contain:

  • _meta.json
  • SKILL.md or lowercase skill.md

The conversion process:

  • maps each OpenClaw skill to one Claude plugin
  • normalizes lowercase skill.md to SKILL.md
  • extracts Claude plugin root assets like agents/, hooks/, .mcp.json, .lsp.json, and settings.json
  • preserves nested skills/ trees when a source package already behaves like a multi-skill plugin
  • skips malformed inputs and records them in reports/generate-marketplace.json

Regenerate The Entire Marketplace

To rebuild the full marketplace from the openclaw-skills submodule:

mise run generate

That regenerates:

  • .claude-plugin/marketplace.json
  • plugins/<plugin-id>/...
  • reports/generate-marketplace.json

The generator replaces previous generated output under .claude-plugin/, plugins/, and reports/ on each run.

If you want the explicit underlying task name, this still works too:

mise run generate_marketplace

For small development runs:

python3 scripts/generate_marketplace.py --source openclaw-skills/skills --output . --limit 10

Validate And Test

Run the generator tests:

mise run test

Validate the generated marketplace:

claude plugin validate .

Notes

  • Plugin ids are generated as <owner>--<slug> after sanitization.
  • Nested source skills/ directories are preserved as plugin content rather than treated as separate top-level source skills.
  • The upstream OpenClaw source is included as the openclaw-skills submodule.

Trustgrade B

  • passBody integrity

    Whether the stored document is plausibly the kind of file the artifact declares, rather than something fetched by mistake.

  • passType matchnot applicable to this artifact type

    Whether the artifact is really the kind of thing its metadata claims it is.

  • passFreshness

    How long since the source repository was last pushed to.

  • passPrompt injection

    Scans the artifact's own text for instructions aimed at your agent rather than at you.

  • warnLicenseno SPDX license detected

    Whether the source repository declares an SPDX license permissive enough to redistribute.

How the grade is calculated

Each check contributes 0 points when it passes, 1 when it warns, and 2 when it fails. The total maps to a letter:

  • Aevery check passed
  • Bone warning
  • Ctwo warnings
  • Dprompt injection or body integrity failed, or three warnings
  • Fone of those failed, and something else is wrong

These are automated hygiene checks, not a security audit, and not a dependency or vulnerability scan. A grade of A means nothing was flagged — not that the artifact is safe.

Versions

  • git-9d1b368787fc2026-07-31