sk-explore
First-link discovery. Gather repo context and write a structured report that feeds into sk-alternatives or sk-planning.
SKILL.mdskills/sk-explore/SKILL.md
GitHubExplore
When to use
- Starting work when the problem, scope, or repo context is unclear.
- The user asks to explore, research, or understand X in this repo.
- Before
sk-alternativesorsk-planningwhen context is missing.
Goal
Produce a self-contained 0 - EXPLORE.md so the next phase can proceed without re-exploring.
Slug and flow folder
- Derive a short kebab-case slug from the goal (confirm with the user if not provided).
- Create
.agents/flows/sk-<slug>/if needed. - Create or update
RUNBOOK.mdwith all phasespending; mark0asin-progress. - Write
0 - EXPLORE.mdand set row0todonewith a one-line summary. - Add
.agents/flows/sk-*/to.gitignoreif this is the first flow in the repo.
Output
0 - EXPLORE.md includes:
- Goal: restated scope.
- Files and Context: relevant files and inline facts.
- Findings: what exists, what's missing, patterns, constraints, edge cases.
- Assumptions and Open Questions: unknowns with acceptance impact.
- Risks: known blockers or high-risk areas.
- Next Step Recommendation:
sk-alternativesorsk-planningwith a one-line reason.
Steps
- Clarify — restate goal and scope. If ambiguity is high, ask or list assumptions.
- Search — find relevant code, tests, docs, and config.
- Read — inline the relevant snippets or facts.
- Synthesize — summarize current state, constraints, and gaps.
- Recommend —
sk-alternativesif the approach is unclear,sk-planningif it is clear. - Write — save the report and update
RUNBOOK.md.
Do not edit non-report files.