| name | ai-explore |
| description | Codebase-only read-only research dispatcher. Thin wrapper around the ai-explore agent for architecture mapping, dependency tracing, pattern identification, and risk surfacing — dispatch for any question spanning >= 1 file, module, or import edge. Trigger for 'explore the codebase', 'where does X live', 'map this module', 'what depends on Y', 'trace this import chain'. Not for external evidence with citations; use /ai-research instead. |
| effort | cheap |
| argument-hint | [question] |
| tags | ["exploration","research","codebase","architecture","mapping"] |
| model_tier | haiku |
| mirror_family | antigravity-skills |
| generated_by | ai-eng sync |
| canonical_source | .claude/skills/ai-explore/SKILL.md |
| edit_policy | generated-do-not-edit |
Explore
Quick start
/ai-explore "where does the install pipeline run hooks?"
/ai-explore "trace the import chain from cli_factory to the durable repository"
/ai-explore "what files reference the legacy ai_providers schema?"
Workflow
Thin wrapper (D-133-09), §10.1 KISS — fewest moving parts. The
ai-explore agent (.agents/agents/ai-explore.md) owns file-reading +
grep and the structured findings output.
- Capture — take the entire argument as the question.
- Dispatch — invoke the
ai-explore agent with the question.
- Report — pass the agent's structured findings through unchanged.
Output Contract
Agent emits Findings / Dependencies / Risks / Recommendations; the
wrapper returns them verbatim. Pure read-only — never edits code, never
fetches external sources.
Integration
Called by: user directly. Dispatches: ai-explore agent
(.agents/agents/ai-explore.md). See also: /ai-research (external
evidence), /ai-review (LLM review), /ai-build (code changes).
$ARGUMENTS
