Phases are sequential. Do not start a phase before the previous one ships. Each phase has a clear deliverable — something a real user can run.
Goal: Core analysis engine works. No AI yet.
Tasks:
- File scanner with ignore list
- Framework detection (Next.js, Express)
- Import/require parser → dependency graph
- Entry point detection from graph topology
- External service detection from imports
- Project map JSON generation
- MD5 file hashing for cache
devmap init— setup wizarddevmap doctor— diagnostics
Deliverable: devmap analyze outputs accurate project structure
without any AI call. Pure static analysis.
Why no AI yet: Validate that the analysis is accurate before adding AI on top. If the foundation is wrong, AI output will be wrong too.
Goal: Users can understand projects faster with AI interpretation.
Tasks:
- Groq integration with provider abstraction layer
- Prompt templates for analyze and ask
- Context Builder — keyword search + file ranking
- Streaming output
- Retry logic + model fallback
- Token-aware context trimming (max 5 files, max 200 lines each)
- Cache integration — skip AI for unchanged files
- Stale snapshot detection + user prompt
- All error scenarios handled (no raw stack traces)
Deliverable: devmap analyze with AI interpretation.
devmap ask with context-aware answers.
Goal: DevMap generates useful project documentation automatically.
Tasks:
devmap docs— generate structured markdown docs folderdevmap onboard— generate onboarding guide with reading order
Deliverable:
docs/
├── project-overview.md
├── architecture.md
├── onboarding.md
└── api.md
Note: Quality bar here is high. Do not ship until output is genuinely useful — not just filled with AI hallucinations.
Goal: DevMap goes beyond understanding into actionable insights.
Tasks:
devmap flow [module]— narrative flow explanationdevmap deadcode— detect unused files, exports, functions (static analysis first, AI for explanation only)devmap report— project health score with recommendations
Note for deadcode: Use static analysis as the detector, AI only for explaining results. Never rely on AI to detect dead code — accuracy must be near 100% or developers won't trust it.
Goal: Users can choose their preferred AI provider.
Tasks:
- OpenAI adapter (GPT-4o mini as default)
- Gemini adapter (1.5 Flash as default)
- Provider selection in
devmap init - Per-project provider override in
.devmap/config.json - Provider-specific model recommendations
Order: OpenAI first (most requested), Gemini second.
Not planned. Not scheduled. Revisit when Phase 5 ships.
- Local AI / Ollama — run DevMap fully offline
- Hybrid Mode — local for small files, cloud for complex queries
- VS Code Extension — DevMap inside the editor
devmap watch— auto-update snapshot on file savedevmap visual— SVG/PNG architecture diagram- Team features — shared snapshots, team config
- Cloud dashboard — web UI for snapshot history
| Version | Phase | Description |
|---|---|---|
| 0.1.0 | 1 | Static analysis only — no AI |
| 0.2.0 | 1 | devmap init + devmap doctor |
| 1.0.0 | 2 | devmap analyze + devmap ask with AI — first public release |
| 1.1.0 | 2 | Performance improvements, cache optimization |
| 1.2.0 | 2 | Express support solidified |
| 2.0.0 | 3 | devmap docs + devmap onboard |
| 3.0.0 | 4 | devmap deadcode + devmap flow + devmap report |
| 4.0.0 | 5 | OpenAI + Gemini support |