A Claude Code plugin that analyzes your codebase and generates project documentation — only the files that actually apply to your project.
| File | Generated when |
|---|---|
PROJECT_OVERVIEW.md |
Always |
API_COLLECTION.md |
REST, GraphQL, gRPC, or tRPC routes exist |
INFRASTRUCTURE.md |
Docker, Kubernetes, Terraform, or cloud IaC exists |
DATA_MODELS.md |
ORM schemas, database models, or migrations exist |
AUTH_FLOWS.md |
JWT, OAuth, session, or auth middleware exists |
ENVIRONMENT_SETUP.md |
.env.example exists or 3+ env vars referenced |
QUICK_FLOW_WALKTHROUGH.md |
Always |
CRON_JOBS.md |
Scheduler library or scheduled tasks exist |
AGENTS.md |
Always |
/plugin marketplace add fenil210/project-mds
/plugin install project-mds@project-mdsNavigate to your project root in the terminal, start Claude Code, then run:
/project-mds
Claude will analyze your codebase, determine which documentation files apply, and generate them in parallel. All files are written to your project root.
- A
project-analyzeragent scans your codebase and outputs a structured report of what documentation applies. - Based on that report, only the relevant writer agents are spawned in parallel.
- Each writer agent reads actual source files deeply — routes, schemas, configs, worker code — and generates its documentation from real code, not templates.
project-mds/
├── .claude-plugin/
│ └── plugin.json # manifest
├── commands/
│ └── project-mds.md # /project-mds orchestrator
└── agents/
├── project-analyzer/ # scans codebase, decides what to generate
├── overview-writer/ # PROJECT_OVERVIEW.md
├── api-collection-writer/ # API_COLLECTION.md
├── infra-writer/ # INFRASTRUCTURE.md
├── data-models-writer/ # DATA_MODELS.md
├── auth-flows-writer/ # AUTH_FLOWS.md
├── env-setup-writer/ # ENVIRONMENT_SETUP.md
├── quick-flow-writer/ # QUICK_FLOW_WALKTHROUGH.md
├── cron-jobs-writer/ # CRON_JOBS.md
└── agents-doc-writer/ # AGENTS.md
- Claude Code latest version
- Run from your project root directory