Remy-CC

Discover and run tests after code modification. Supports multi-angle defect prediction and coverage analysis. Recommended after large-scale edits or accumulated working tree changes.

Scan source files (Python, C/C++, TypeScript) and generate semantic logic index via LLM. Index is injected into CLAUDE.md and used by hooks/skills for dependency analysis. Run after repo init or major changes.

Deep repository analysis with multi-agent parallel perspectives. Requires /init + /remy-index as prerequisites. Produces structured research reports.

Diagnose bugs via hypothesis loop with circuit breaker. Diagnosis only — produces evidence packets for /remy-patch without modifying code.

Audit architecture for risks, side effects, and ambiguities before writing code. Produces evidence packets for /remy-patch. Recommended for large or complex tasks.

Verify code changes via Intent/Log/Code triangulation. Independent blind audit without prior session context. Paired with /remy-changelog: generate log, rewind context, then audit.

Analyze CI/CD failure logs to diagnose build, test, and gate failures. Supports GitHub Actions (gh CLI), local log files, and pasted logs. Produces evidence packets for /remy-patch.

Apply code modifications under engineering constraints. Accepts plan packets from /remy-plan as argument, or use standalone with a modification request.

Review branch changes for security vulnerabilities. Multi-stage: regex pre-scan, parallel agents, and false-positive filtering.

Generate persistent unit tests for existing or stub code. Supports post-hoc testing (default) and TDD mode (--tdd). Multi-angle agent analysis at medium/high effort levels.

Display the current logic index tree (.claude/logic_tree_view.md) without regenerating.

Analyze a GitHub repository in two stages: metadata assessment via GH CLI, then sandboxed deep inspection upon confirmation.

Generate a temporary change log recording code modifications. Paired with /remy-audit: generate log, rewind AI context (preserving code changes), then run blind audit.

Record technical decisions and progress as a persistent milestone report. Use before /compact or at task completion.

Refresh the project directory tree snapshot (.claude/project_tree.md). Use after file operations that alter structure.