All authors
Claude Skills by sunitghub
github.com/sunitghub25 skills0 installs1 views
- Ai AuditAudits an AI/LLM codebase across nine surfaces using the SCAN method and returns a ship/conditional/hold verdict. Use when asked to review, audit, or security-check an AI agent, LLM app, RAG pipeline, or prompt/tool-calling system for AI-specific risks. Static analysis only.Votes: 0GitHub stars: 9
- Bulk ReadSize-gated citation-only reading for large files — checks line count, reads directly under the threshold, dispatches a fresh-context Haiku subagent that returns only file:line citations over it. Use when exploring/understanding a large file (not editing it) in any sprint tier, not just high-risk orient research.Votes: 0GitHub stars: 9
- Canon SlidesGenerates Marp slide decks from canon knowledge. Use when asked to create slides, build a presentation, or generate a deck on a canon topic (context management, evaluator pattern, or skill authoring). Renders to HTML and PPTX for browser viewing and SharePoint/PowerPoint distribution.Votes: 0GitHub stars: 9
- CaptureRecords non-obvious discoveries, constraints, and gotchas to HANDOFF.md. Use when something surprising is found mid-sprint.Votes: 0GitHub stars: 9
- Context CheckAudits always-on context load for bloat, redundancy, and quality. Use when context feels heavy or periodically to keep the always-on budget lean.Votes: 0GitHub stars: 9
- Context DoctorAudits a repo's agent context — system prompt, CLAUDE.md/AGENTS.md, skills, and references — against context-engineering lessons for Claude 4/5 models (model detected from the running session, or set via --model), then writes claude-optimization.md with a Summary table. Use to right-size an agent setup, cutting over-constraint, redundancy, always-upfront context, and conflicting instructions.Votes: 0GitHub stars: 9
- Dead Code CleanupScans a repo for likely-unreferenced top-level symbols (JS/TS exports, Python def/class, Go exported func), reports them as removal candidates, and writes a structured `dead-code-report.md` (summary + high/low-confidence candidate tables). Use when asked to find dead code, unused exports, or unreferenced functions, or to clean up a codebase. Advisory-first — never deletes without explicit confirmation.Votes: 0GitHub stars: 9
- Doc AuditAudit user-facing docs for overstated claims, missing prerequisites, absolutes, scope inflation, and stale commands.Votes: 0GitHub stars: 9
- Learnings SweepAggregates per-ticket UNPROMOTED .tickets/<id>/learnings.md candidates into a single capped root LEARNINGS.md index, without promoting any of them. Use after `tkt learn <id>` confirms a candidate (single-ticket mode, called from sprint complete), or run `--full` by hand to backfill/reconcile the whole repo.Votes: 0GitHub stars: 9
- MikadoMaps the dependency tree for a refactor before changing code, using the Mikado Method — attempt the goal, revert on breakage, record prerequisites, execute leaves-first. Use when a refactor is likely to cascade into prerequisite changes or touch interdependent modules.Votes: 0GitHub stars: 9
- Mutation TestApplies small logic mutations to a sprint's changed logic files and asserts the test suite fails on each — a surviving mutant is a test that cannot fail. Use when asked to mutation-test, check whether tests have teeth, or find tests that pass no matter what the code does. Advisory, not a close gate.Votes: 0GitHub stars: 9
- Output ValidatorValidates agent-generated reports and summaries before delivery. Catches generator-evaluator collapse — where the AI summarizes data without checking if the summary is true. Run before delivering any report, status update, or data summary.Votes: 0GitHub stars: 9
- Pr ReviewReviews open PRs on sunitghub/canon-skills — fetches diffs, presents findings, updates PRs/canon-skills-pr-review.md, creates tickets for Fix=Y PRs, and posts approved comments. Use when the user asks to review canon-skills PRs or check what's open on canon-skills.Votes: 0GitHub stars: 9
- Promote LearningsReviews LEARNINGS.md's UNPROMOTED rows as a fresh, no-implementation-history reader and proposes where each durable one belongs. Report-only, except that in a consumer project it writes the proposals you confirm to PROMOTED.md. Use as one of Upkeep's four report-only checks, or by hand when the UNPROMOTED queue needs triage.Votes: 0GitHub stars: 9
- Repo AuditAudit a repo across four dimensions — uniqueness and selling point, philosophy-to-implementation coherence, setup and doc effectiveness, and codebase quality (DRY, KISS, SRP, composition). Invoke when asked to review, assess, or health-check a repo.Votes: 0GitHub stars: 9
- Repo Workflow AuditRuns a fresh-context, 4-way parallel audit of a canon workflow's own skill/reference docs against themselves — pipeline gates, adversarial review, cross-doc consistency, stale-reference sweep. Use when asked to deep-scan, audit, or health-check the sprint/wrapup workflow docs (or another skill's own protocol files) for internal contradictions, staleness, or drift — not for auditing a whole repo's product code (use repo-audit for that).Votes: 0GitHub stars: 9
- Sdlc AuditAudits a repo's development workflow against Anthropic's AI-Native SDLC Playbook (twenty practices across plan, design, build, test, deploy, maintain) and rates each Covered, Partial, or Gap with file:line evidence. Use when asked to check a repo's dev process maturity, compare a workflow to the AI-native SDLC playbook, or assess how "AI-native" a team's build/test/deploy pipeline is.Votes: 0GitHub stars: 9
- Skill EvalRuns execution evals for a named skill against test cases in evals/evals.json. Use when you want to verify a skill produces correct output for known prompts, check skill quality after edits, or confirm a new skill works before registering it.Votes: 0GitHub stars: 9
- Skill ExportExports any flat canon skill as a paste-ready prompt, or refines an existing prompt using efficiency.md standards. Use when running a skill in claude.ai, a system prompt, or another environment that doesn't load SKILL.md natively. Invoke as skill-export <skill-name> or skill-export <skill-name> "<prompt>". Rejects skills with sub-skills.Votes: 0GitHub stars: 9
- SprintManages the sprint workflow for focused changes. Use when asked to add, fix, update, implement, debug, or build — see the Workflow tiers section for what's out of scope.Votes: 0GitHub stars: 9
- WrapupInternal close-pipeline invoked by sprint complete — quality checks, code review, cleanup. Not for direct use; a normal/high-risk sprint's own "sprint complete" step calls this automatically.Votes: 0GitHub stars: 9
- Bad FrontmatterThe '---' block must be the first line.Votes: 0GitHub stars: 9
- GoodFormats commit messages. Use when the user asks for a commit message.Votes: 0GitHub stars: 9
- HooksRegisters a hook and pushes code.Votes: 0GitHub stars: 9
- No EvalsA skill with no evals folder.Votes: 0GitHub stars: 9