This bundle turns the attached Heady directives and configuration seeds into a production-oriented monorepo with Cloudflare edge workers, Cloud Run services, shared runtime packages, pgvector migrations, deployment workflows, and operational runbooks. Internal source material comes from the attached workspace directives, HCFullPipeline definitions, and cognitive configuration.
Cloudflare Vectorize supports up to 5 million vectors per index and up to 1536 dimensions, which makes it a strong edge retrieval layer while origin pgvector remains the authoritative memory plane (Cloudflare Vectorize).
HeadyMCP publicly positions itself as an edge-native MCP server with JSON-RPC transport, SSE support, and 30+ tools for IDEs such as VS Code, Cursor, and Windsurf (HeadyMCP).
HeadyAPI publicly positions itself as a liquid gateway that races 4+ providers and uses auto-failover, which directly informs the worker routing layer included here (HeadyAPI).
- Shared phi and CSL packages
- 15 Cloud Run service scaffolds
- 4 Cloudflare worker scaffolds
- pgvector and Graph RAG migrations
- CI/CD, Docker, smoke tests, and deployment templates
- Reconciled directives and missing cognitive-layer files
- Runbooks, ADRs, and activation checklists
- Install
pnpmand Node 22. - Copy
.env.exampleto.envand fill secrets. - Provision Cloudflare, GCP, PostgreSQL with pgvector, and Redis.
- Run
pnpm installthenpnpm build. - Run
pnpm devordocker compose up --build. - Deploy Cloudflare workers and Cloud Run services via the included GitHub Actions.
| Guide | What You Learn | Time |
|---|---|---|
| HeadyBuddy | AI companion (desktop + mobile) | 15 min |
| HeadyServices | Backend infrastructure | 10 min |
| HeadyAI-IDE | Custom IDE setup | 20 min |
| HeadyBrowser | Browser extension | 10 min |
| HeadyMCP | Manager Control Plane | 15 min |
| Heady API | API interaction | 10 min |
- Ecosystem Overview - Full system map across all brands
- Service Integration - How Heady services connect
- HeadyManager API - Service endpoints reference
heady-manager.js # Node.js API Gateway (port 3300)
├── src/hc_pipeline.js # HCFullPipeline engine
├── src/agents/ # Builder, Researcher, Claude Code, Deployer
├── packages/hc-supervisor # Multi-agent Supervisor (parallel fan-out)
├── packages/hc-brain/ # HeadyBrain meta-controller
├── packages/hc-health/ # Health checks + cron
├── configs/ # YAML configs (source of truth)
├── public/ # Static frontend (this site)
│ ├── index.html # Landing page
│ ├── docs.html # Documentation hub
│ ├── products.html # Product catalog
│ ├── api-docs.html # Interactive API docs
│ ├── status.html # System status
│ └── verticals/ # Product vertical pages
├── frontend/ # React admin IDE (Vite)
├── scripts/ # Build, deploy, checkpoint automation
├── docs/ # Developer documentation
└── heady-registry.json # Central component catalog
Five-stage deterministic execution with deep checkpoints:
ingest → plan → execute-major-phase → recover → finalize
Operational Readiness Score (ORS) gates execution intensity:
- 85–100: Full parallelism, all optimisations enabled
- 70–85: Normal operation
- 50–70: Maintenance mode, no new large builds
- <50: Recovery only — repair before building
| Endpoint | Method | Description |
|---|---|---|
/api/health |
GET | Health check |
/api/system/status |
GET | Full system status |
/api/pipeline/run |
POST | Trigger pipeline run |
/api/pipeline/state |
GET | Current state |
/api/supervisor/status |
GET | All agent statuses |
/api/brain/status |
GET | Brain + ORS |
/api/registry |
GET | Component registry |
Full reference: public/api-docs.html or /api-docs.html on the running system.
| Variable | Required | Description |
|---|---|---|
ANTHROPIC_API_KEY |
Yes | Claude API key |
DATABASE_URL |
Recommended | PostgreSQL connection |
HEADY_API_KEY |
Recommended | Auto-generated if not set |
PORT |
Optional | Server port (default: 3300) |
NODE_ENV |
Optional | development or production |
The Heady CLI provides command-line access to Heady services:
# Set API key (or add to .env)
export HEADY_API_KEY="your_api_key"
# Run CLI
npm run cli
# or directly:
python scripts/heady_cli.pySee scripts/heady_cli.md for full documentation.
At every checkpoint, 10-step deep analysis detects drift and syncs all files:
.\scripts\checkpoint-sync.ps1 # Full sync
.\scripts\checkpoint-sync.ps1 -Mode check # Read-only drift detection
.\scripts\checkpoint-sync.ps1 -Mode fix # Auto-fixFull protocol: docs/CHECKPOINT_PROTOCOL.md
| Path | Purpose |
|---|---|
public/docs.html |
Documentation hub |
CLAUDE.md |
Claude Code integration protocol |
docs/CHECKPOINT_PROTOCOL.md |
Sync protocol |
docs/C4_ARCHITECTURE.md |
Architecture diagrams |
configs/hcfullpipeline.yaml |
Pipeline definition (source of truth) |
heady-registry.json |
Central component catalog |
This project supports Python development with PyCharm and Google Colab integration.
data/ # Data directories
├── raw/ # Raw data files (git-ignored)
├── processed/ # Processed data
└── external/ # External data (git-ignored)
notebooks/ # Jupyter notebooks
├── exploratory/ # Scratch, EDA, experiments
├── reports/ # Clean, final notebooks
├── archive/ # Retired notebooks
└── figures/ # Exported plots/images
src/ # Reusable Python code
tests/ # Unit tests
# Install Python dependencies
pip install -r requirements.txt
# For PyCharm: Open as Python project
# For Colab: See COLAB_WORKFLOW.mdrequirements.txt- Python dependenciesNOTEBOOK_TEMPLATE.md- Template for new notebooksCOLAB_WORKFLOW.md- Google Colab workflow guide.env.example- Environment variables template
.\commit_and_build.ps1 # Local build
.\nexus_deploy.ps1 # Push to all remotesCloud Run deploy: npm run deploy:cloud-run
Proprietary — Heady Systems