a686d28669
Phase 05: Subagent Codebase Analysis - 05-03: gsd-indexer subagent for file reading and export/import extraction - 05-04: Refactor analyze-codebase Steps 2-3 to use indexer subagent Gap closure addresses context exhaustion during indexing phase (Steps 2-3), not just entity generation (Step 9). Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
5.5 KiB
5.5 KiB
Roadmap: v1.9.0 Codebase Intelligence System
Goal: Make GSD feel intelligent and automagical in how it navigates and understands both greenfield and brownfield projects.
Phases: 5 (4 complete)
Current Milestone: v1.9.0
Phase 1: Foundation & Learning ✓
Goal: Establish index schema and incremental learning via PostToolUse hook Status: Complete Plans: 2/2
Phase 2: Context Injection ✓
Goal: Inject codebase awareness into every session via SessionStart hook Status: Complete Plans: 2/2
Phase 3: Brownfield & Integration ✓
Goal: Deep analysis command for existing codebases, workflow integration Status: Complete Plans: 3/3
Phase 4: Semantic Intelligence & Scale ✓
Goal: Transform syntax-only indexing into semantic understanding with graph-based relationships Status: Complete Plans: 5/5
Plans:
- 04-01-PLAN.md — SQLite graph layer with sql.js (Wave 1)
- 04-02-PLAN.md — Graph-backed rich summary generation (Wave 2)
- 04-03-PLAN.md — Semantic entity generation via Claude API (Wave 2)
- 04-04-PLAN.md — CLI query interface for getDependents (Wave 3)
- 04-05-PLAN.md — Wire plan-phase.md to inject intel into planner (Wave 3)
Wave Structure:
- Wave 1: 04-01 (SQLite foundation)
- Wave 2: 04-02, 04-03 (parallel - both depend only on 04-01)
- Wave 3: 04-04, 04-05 (parallel - consumption layer)
Why this phase:
- Current system provides "2-3 ls commands worth of information" (Claude's own assessment)
- Missing: what files actually DO, who uses them, blast radius of changes
- Senior engineers at top companies need real intelligence, not file counts
Delivers:
- SQLite graph layer (sql.js - zero native deps) for relationship queries
- Entity-based semantic documentation (Claude writes understanding, not just syntax)
- Semantic
/gsd:analyze-codebasethat creates initial entities - Rich summary generation from accumulated semantic knowledge
- CLI query interface for "what uses this file?" queries
Requirements:
- INTEL-04: Entity files capture semantic understanding (purpose, what exports do)
- INTEL-05: Relationships queryable ("what uses this file?", "blast radius")
- INTEL-06:
/gsd:analyze-codebasecreates initial entity docs via Claude - INTEL-07: Summary reflects accumulated semantic knowledge
Success Criteria:
- Claude can answer "what uses src/lib/db.ts?" from SessionStart context
- Summary includes file purposes, not just file counts
- Transitive dependency queries work (blast radius)
- Works at scale (500+ file codebases)
Phase 5: Subagent Codebase Analysis
Goal: Prevent context exhaustion on large codebases by delegating analysis to subagents Depends on: Phase 4 Status: Gap closure in progress Plans: 4 plans (2 complete, 2 gap closure)
Plans:
- 05-01-PLAN.md — Create gsd-entity-generator subagent (Wave 1)
- 05-02-PLAN.md — Refactor Step 9 for subagent delegation (Wave 2) — partial, gap found
- 05-03-PLAN.md — Create gsd-indexer subagent for Steps 2-3 (Wave 3) — gap closure
- 05-04-PLAN.md — Refactor Steps 2-3 for subagent delegation (Wave 4) — gap closure
Wave Structure:
- Wave 1: 05-01 (entity generator agent)
- Wave 2: 05-02 (Step 9 refactor)
- Wave 3: 05-03 (indexer agent) — gap closure
- Wave 4: 05-04 (Steps 2-3 refactor + verification) — gap closure
Why this phase:
- Current entity generation loads file contents in orchestrator context
- On large codebases (500+ files), orchestrator exhausts context during file selection and batching
- Subagent delegation gives fresh 200k context for file processing
Gap Found (05-VERIFICATION.md):
- Original scope only addressed Step 9 (entity generation)
- Actual context exhaustion occurs during Steps 2-3 (indexing)
- Orchestrator reads ALL file contents during indexing, not just entity generation
- Need additional gsd-indexer subagent for Steps 2-3
Delivers:
gsd-entity-generatorsubagent following gsd-codebase-mapper patterngsd-indexersubagent for file reading and export/import extraction- Refactored
/gsd:analyze-codebasewith full subagent delegation - Preserved orchestrator context for large codebase analysis
Requirements:
- INTEL-08: Entity generation delegated to subagent (not inline) ✓
- INTEL-09: Subagent writes entities directly, returns statistics only ✓
- INTEL-10: Orchestrator passes file paths, not file contents — BLOCKED (needs 05-03, 05-04)
- INTEL-11: Indexing phase delegated to subagent (gap closure)
Success Criteria:
- Entity generation works via subagent spawn ✓
- Indexing works via subagent spawn (gap closure)
- Orchestrator context preserved (no file contents loaded in orchestrator)
- Entities correctly formatted and graph.db updated ✓
- Works on 500+ file codebases without context exhaustion
Traceability
| Requirement | Phase | Status |
|---|---|---|
| INTEL-01 | Phase 1 | ✓ Complete |
| INTEL-02 | Phase 2 | ✓ Complete |
| INTEL-03 | Phase 3 | ✓ Complete |
| INTEL-04 | Phase 4 | ✓ Complete (04-03) |
| INTEL-05 | Phase 4 | ✓ Complete (04-04) |
| INTEL-06 | Phase 4 | ✓ Complete (04-03) |
| INTEL-07 | Phase 4 | ✓ Complete (04-02) |
| INTEL-08 | Phase 5 | ✓ Complete (05-01, 05-02) |
| INTEL-09 | Phase 5 | ✓ Complete (05-01) |
| INTEL-10 | Phase 5 | Gap closure (05-03, 05-04) |
| INTEL-11 | Phase 5 | Gap closure (05-03, 05-04) |
Created: 2026-01-19 Updated: 2026-01-20 — Phase 5 gap closure plans added (05-03, 05-04)