my-fetch-repo

name	my-fetch-repo
description	Use when you need to fetch a GitHub repo's metadata, README, and file tree — wraps gh CLI. Returns structured data. Also use for "check this repo", "what is this repo", or "fetch repo info".
argument-hint	< owner/repo | GitHub URL >

/my-fetch-repo — GitHub Repository Fetcher

Fetches repo metadata, README content, and file tree using gh CLI. Returns structured data for standalone use or as input to /my-research-targets pipeline.

Prerequisites

• gh CLI installed and authenticated (gh auth status)

Usage

• /my-fetch-repo https://github.com/owner/repo — fetch and display
• /my-fetch-repo owner/repo — fetch by owner/repo
• Called by /my-research-targets for GitHub URLs

Fetch Strategy

Step 1 — Metadata (always)

gh repo view <owner>/<repo> --json name,description,stargazerCount,forkCount,licenseInfo,primaryLanguage,topics,url,createdAt,updatedAt

Step 2 — README (always)

gh api repos/<owner>/<repo>/readme --jq '.content' | base64 -d | head -200

Step 3 — File tree (for repos with >100 stars)

gh api 'repos/<owner>/<repo>/git/trees/main?recursive=1' --jq '.tree[] | select(.type=="blob") | .path' | head -80

Fallback

If gh fails (rate limit, private repo):

WebFetch "https://github.com/<owner>/<repo>" with prompt "Extract repo description, stars, language, and README summary"

Output

## owner/repo
Stars: 12,345 | License: MIT | Language: TypeScript
Topics: agent, cli, mcp
Created: 2025-06-15 | Updated: 2026-03-18

### README (first 200 lines)
<readme content>

### File Tree (top 80 files)
<file paths>

When called by /my-research-targets, return raw data for classification.

Gotchas

• gh api has rate limits (5000/hr authenticated) — batch requests when fetching multiple repos
• File tree endpoint returns main branch by default — some repos use master; fallback: gh api repos/<owner>/<repo> --jq '.default_branch'
• Large READMEs (>200 lines) should be truncated — head -200 is sufficient for classification
• Private repos return 404 — check access before reporting as "not found"

Rules

• Read-only — never star, fork, create issues, or modify anything
• Always check gh auth status before first use in a session
• Truncate README to 200 lines and file tree to 80 files to stay within context budget