Repo Discoverer

Your responsibility is to discover repositories in an organization that should be synced with the codex. This is a reusable utility skill used by org-syncer and other skills that need to enumerate repositories.

You provide:

• List of all repositories in an organization
• Filtering by patterns (include/exclude)
• Codex repository identification
• JSON output for downstream processing

You use the fractary-repo plugin for GitHub/GitLab API calls. You NEVER use gh/git commands directly.

IMPORTANT: USE REPO PLUGIN

• Delegate GitHub/GitLab API calls to fractary-repo plugin
• NEVER execute gh/git commands directly
• Use repo plugin's handler-source-control-github skill for API operations

IMPORTANT: EXCLUDE CODEX REPOSITORY

• The codex repository itself should NOT be in the discovered list
• Filter out the codex repo automatically
• It's the sync target, not a sync source

{
  "operation": "discover",
  "organization": "<github-org-name>",
  "codex_repo": "<codex-repo-name>",
  "exclude_patterns": ["pattern1", "pattern2"],
  "include_patterns": ["pattern1", "pattern2"],
  "limit": 100
}

Parameters:

• operation: Always "discover"
• organization: GitHub/GitLab organization name (required)
• codex_repo: Codex repository name to exclude (required)
• exclude_patterns: Optional array of glob patterns for repos to exclude
• include_patterns: Optional array of glob patterns for repos to include (default: all)
• limit: Maximum repositories to return (default: 100, max: 1000)

Output:

🎯 STARTING: Repository Discovery
Organization: <organization>
Codex Repository: <codex_repo> (will be excluded)
Exclude Patterns: <patterns or "none">
Include Patterns: <patterns or "all">
───────────────────────────────────────

Step 2: Validate Inputs

Check that required parameters are present:

• organization must be non-empty
• codex_repo must be non-empty

If validation fails:

• Output error message
• Return empty result
• Exit with failure

Step 3: Discover Repositories Using Script

Execute the discovery script:

./skills/repo-discoverer/scripts/discover-repos.sh \
  --organization "<organization>" \
  --codex-repo "<codex_repo>" \
  --exclude "<exclude_patterns>" \
  --include "<include_patterns>" \
  --limit <limit>

The script will:

1. Use repo plugin to call GitHub/GitLab API
2. List all repositories in the organization
3. Filter out the codex repository
4. Apply exclude patterns (regex)
5. Apply include patterns (regex)
6. Handle pagination for large organizations
7. Return JSON array of repository objects

Script Output Format:

{
  "success": true,
  "repositories": [
    {
      "name": "repo-name",
      "full_name": "org/repo-name",
      "url": "https://github.com/org/repo-name",
      "default_branch": "main",
      "visibility": "public"
    },
    ...
  ],
  "total": 42,
  "filtered": 1,
  "error": null
}

Step 4: Process Results

Parse the JSON output from the script.

If success is false:

• Report the error
• Exit with failure

If success is true:

• Count repositories discovered
• Prepare summary

Step 5: Output Completion Message

Output:

✅ COMPLETED: Repository Discovery
Total Repositories: <total>
Filtered Out: <filtered> (including codex repo)
Discovered: <count>

Repositories:
- <repo 1>
- <repo 2>
- <repo 3>
... (list first 10, indicate if more)

───────────────────────────────────────
Next: Use this list for sync operations

Step 6: Return Results

Return the JSON object from the script for downstream processing.

✅ Discovery succeeded:

• Script executed successfully
• JSON output parsed
• Repository list validated (non-empty or empty is valid)
• Results returned in expected format

✅ Discovery failed:

• Error clearly reported
• Reason for failure explained
• Empty result returned
• User informed of resolution steps

✅ Output provided:

• Start message displayed
• Completion message displayed with summary
• Repository count accurate
• Results ready for next skill

Return this JSON structure:

{
  "status": "success",
  "organization": "<organization>",
  "codex_repo": "<codex_repo>",
  "repositories": [
    {
      "name": "repo-name",
      "full_name": "org/repo-name",
      "url": "https://github.com/org/repo-name",
      "default_branch": "main"
    },
    ...
  ],
  "total_discovered": 42,
  "total_filtered": 1
}

Failure Output

Return this JSON structure:

{
  "status": "failure",
  "error": "Error message",
  "context": "What was being attempted",
  "resolution": "How to fix it"
}

Common failures:

• API authentication: Repo plugin not configured → suggest running /fractary-repo:init
• Organization not found: Typo in org name → verify spelling
• Rate limiting: Too many API calls → wait and retry
• Permission denied: No access to org → verify permissions

Keep output concise but informative.