| name | issue-scout |
| description | Find good issues to work on by scanning repos for approachable tasks |
| user_invocable | true |
| always_enabled | true |
Issue Scout Process
When asked to find good issues to work on, follow these steps:
1. Search for Approachable Issues
Scan target repositories for issues labeled as good entry points:
gh issue list --repo <repo> --label "good first issue" --state open --json number,title,labels,createdAt,url,body
gh issue list --repo <repo> --label "help wanted" --state open --json number,title,labels,createdAt,url,body
If no repo is specified, search across an organization:
gh search issues --owner <org> --label "good first issue" --state open --sort updated --json repository,number,title,labels,url
gh search issues --owner <org> --label "help wanted" --state open --sort updated --json repository,number,title,labels,url
2. Filter by Language or Area
If the user specifies a language or area of interest, narrow results:
gh search issues --owner <org> --label "good first issue" --language <language> --state open --json repository,number,title,labels,url
Also check for area-specific labels like area/cli, area/api, kind/documentation, kind/bug, kind/feature, or similar conventions used by the target repos.
3. Check Issue Freshness
For each candidate issue, verify it is still actionable:
- Is the issue already assigned? If so, skip it unless the assignment is stale (no activity for 30+ days).
- Has someone already opened a PR for it? Check linked PRs.
- Has there been recent discussion indicating the approach is unclear or blocked?
- Is the issue still relevant to the current codebase version?
gh issue view <number> --repo <repo> --json assignees,comments,linkedPullRequests
4. Rank by Estimated Effort
Classify each issue by effort:
- Quick win (< 1 hour): Typo fixes, small config changes, simple bug fixes with clear reproduction steps
- Half day (1-4 hours): Single-component changes, adding tests, minor feature additions
- Full day (4-8 hours): Multi-file changes, new features requiring some design, integration work
Ranking factors:
- Clarity of the issue description (clear > vague)
- Availability of reproduction steps or test cases
- Scope of changes needed (fewer files > more files)
- Prior discussion that clarifies the approach
- Recency (recently filed issues are more likely to be relevant)
5. Provide Context
For the top recommended issues, include helpful context:
- Links to relevant source files the contributor will need to understand
- Pointers to related tests they can use as examples
- Any comments from maintainers that clarify expectations
- Links to contributing guides or development setup docs
6. Output Format
Structure the recommendations as:
Recommended Issues:
For each issue (ordered by best fit first):
#: (<repo>)
- Effort: Quick win / Half day / Full day
- Labels:
- Why: Brief explanation of why this is a good pick
- Context: Pointers to relevant code or docs
- Link:
Filters Applied: Language, area, or other criteria used
Total Found: X issues scanned, Y recommended
Limit recommendations to the top 5-10 most actionable issues. Prioritize issues that are clearly scoped, unassigned, and have maintainer input clarifying the expected approach.