semantic-wake-intelligence

This MCP server implements the Wake Intelligence brain, a 3-layer temporal intelligence system that combines causality (Past), memory management (Present), and predictive pre-fetching (Future) to enhance AI agent decision making. The server exposes tools for tracking WHY contexts were created, managing active vs. archived data, and predicting WHAT contexts will be needed next. You can use it to trace decision history, manage memory tiers, and proactively fetch relevant contexts. To run it locally, install the CLI tooling (wrangler) and deploy the Cloudflare workers-based MCP using the provided configuration. The available capabilities include causal-chain tracking, 4-tier memory classification, LRU-based prioritization, and composite prediction scoring for pre-fetching. The architecture is hexagonal, enabling you to test business logic independently from infrastructure, while preserving semantic intent across layers.

Prerequisites:

• Node.js 20.x or higher
• npm (comes with Node.js)
• Wrangler CLI (for Cloudflare Workers) - install with: npm install -g wrangler

Installation steps:

1. Clone the repository: git clone https://github.com/semanticintent/semantic-wake-intelligence-mcp.git cd semantic-wake-intelligence-mcp

2. Install dependencies: npm install

3. If you plan to deploy to Cloudflare Workers, configure Wrangler and D1:

   • Copy example Wrangler config if provided, or create wrangler.jsonc
   • Create a D1 database for the MCP: wrangler d1 create mcp-context
   • Update wrangler.jsonc with the generated database_id

4. Run local migrations and start the server (as described in the repo’s quick start): wrangler d1 execute mcp-context --local --file=./migrations/0001_initial_schema.sql wrangler d1 execute mcp-context --file=./migrations/

5. Start or deploy the MCP according to your environment preferences (local dev vs. production):

   • Local development: wrangler dev
   • Production deploy: wrangler publish

Note: The MCP uses a Cloudflare Workers-based runtime with a D1 database for persistence. Ensure your environment variables (e.g., D1 database_id) are configured as described in the repository documentation.

Tips and caveats:

• Ensure Node.js 20.x is installed to match the server’s TypeScript target.
• If migrating from a previous setup, keep your D1 migrations under migrations/ and run them in order.
• The 3-layer architecture (Past/Present/Future) relies on time-based scoring; consider adjusting decay rates only after profiling in your workload.
• For deployment, the Wrangler CLI requires authentication with Cloudflare; run wrangler login before first deploy.
• If you see memory issues, adjust Node.js memory limit via NODE_OPTIONS in your environment variables.
• The MCP relies on semantic contracts; maintain clear naming for contexts and events to preserve observability.