Agent skill
github-repo-sanity
Sanity-check GitHub repositories before citing, recommending, or comparing them. Use when an agent refers to GitHub repos, OSS libraries, starter kits, templates, SDKs, MCP servers, or example projects. Verify repo health using recent commit activity and adoption signals such as stars before recommending it. Prefer active repos; explicitly flag stale, archived, or low-signal repos instead of presenting them as good defaults.
Install this agent skill to your Project
npx add-skill https://github.com/inkeep/agents/tree/main/.agents/skills/github-repo-sanity
SKILL.md
GitHub Repo Sanity
When mentioning GitHub repositories, do not treat existence as endorsement.
Default behavior: if a repo looks stale, weakly adopted, or unmaintained, avoid recommending it as a default option.
When to use this skill
Use this skill when you:
- Recommend GitHub repositories to a user
- Compare multiple repositories or OSS options
- Cite a repo as an implementation model or best-in-class example
- Suggest a starter, template, SDK, tool, integration, or MCP server hosted on GitHub
Do not use this skill when the repo is only being mentioned as historical context, a private internal dependency, or a direct answer to "what is this repo?" In those cases, still flag staleness if it matters.
Required checks
Before recommending a repo, verify:
- Whether the repo is archived
- How recent the latest meaningful commit is
- The repo's star count
- Whether the repo still appears maintained enough for the user's purpose
If browsing is available, verify using the live GitHub repo page or equivalent primary source. Do not guess on recency or stars.
If you cannot verify live repo health, do not present the repo as a confident default recommendation. Either mark it as unverified or prefer a repo you can verify.
Decision rule
Treat recent activity as the primary signal and stars as a secondary sanity check.
Recommend
Safe to recommend by default when most of the following are true:
- Not archived
- Latest meaningful commit is recent enough for the category
- Star count is credible for the use case or at least not suspiciously low
- Nothing else suggests abandonment
Mention with caveat
Mention, but explicitly qualify, when:
- The repo is somewhat stale but still relevant
- Stars are low for a repo being presented as a general-purpose default
- The repo may be good for a narrow use case, but not a broadly safe recommendation
Avoid as a default recommendation
Do not present the repo as a strong option when any of these are true:
- Archived
- No meaningful activity for a long period
- Very low stars combined with stale activity
- Better-maintained alternatives are available
Practical thresholds
Use judgment by category, but default to these heuristics:
- Green: activity within roughly the last 6 months
- Yellow: last activity roughly 6 to 18 months ago
- Red: last activity older than roughly 18 months
Adjust those expectations by category:
- Fast-moving tools: AI SDKs, starters, templates, and agent frameworks should usually show recent activity
- Stable infrastructure: Mature libraries, protocol implementations, and low-churn utilities may still be healthy with slower commit cadence
- Reference repos: Example apps and demos should not be treated as strong defaults unless they also show recent maintenance
Stars are relative to the ecosystem and use case. Use these only as rough anchors:
- Strong signal: clearly well adopted relative to similar repos in the same ecosystem
- Moderate signal: enough adoption to show real usage, but not necessarily category-leading
- Weak signal: low adoption relative to peers, especially when combined with stale activity
Do not reject a niche repo only because it has few stars. Low stars matter most when the repo is being proposed as a broadly trustworthy default and better-maintained alternatives exist.
Response requirements
When recommending a repo, include short evidence:
- Repo name with link
- Latest activity date
- Star count
- A brief verdict:
recommend,caution, oravoid
Use concrete dates, not vague phrases like "recent" or "old."
Important nuance
- Historical context: If the user explicitly wants older or historically important repos, you may include them, but say they are historical references rather than current recommendations.
- Popularity is not maintenance: If a repo is widely starred but clearly stale, say so. Popularity does not override maintenance risk.
- New but promising: If a repo is new but fast-moving and already has credible adoption, it can still be a valid recommendation.
- Unverified is not endorsed: If you cannot verify live repo health, say that the recommendation is unverified instead of presenting it confidently.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
structured-itinerary-responses
Present time-aware itineraries with clear actions and citations
weather-safety-guardrails
Keep activity suggestions safe and respect local conditions
api-logging-guidelines
Best practices and guidelines for using logger in API routes. Defines appropriate logging levels, what to log, and when to avoid logging. Use when implementing or reviewing API route logging, debugging strategies, or optimizing log output.
vercel-composition-patterns
React composition patterns that scale. Use when refactoring components with boolean prop proliferation, building flexible component libraries, or designing reusable APIs. Triggers on tasks involving compound components, render props, context providers, or component architecture. Includes React 19 API changes.
write-docs
Write or update documentation for the Inkeep docs site (agents-docs package). Use when: creating new docs, modifying existing docs, introducing features that need documentation, touching MDX files in agents-docs/content/. Triggers on: docs, documentation, MDX, agents-docs, write docs, update docs, add page, new tutorial, API reference, integration guide.
product-surface-areas
Consolidated dependency graph of Inkeep customer-facing surface areas (UIs, CLIs, SDKs, APIs, protocols, config formats). Example use: as a prd-time (planning/brainstorming phase) or post-change checklist to understand the full scope of side-effects or what making one change to the product means for the rest. Use whenever you need to understand the "ripple" out effects of any change.
Didn't find tool you were looking for?