How coryhacking/wavefoundry works
Wavefoundry is a spec-driven development framework for AI coding agents, combining a local MCP tool server with repository indexing and workflow support. The documented setup flow bootstraps the framework, builds a project index, configures the agent host, and hands control back after an MCP restart. The README excerpt establishes setup and indexing relationships; internal links among lifecycle, memory, and retrieval modules are not shown, so they are omitted rather than inferred.
Main entry points: setup_wavefoundry.py Setup Orchestrator · indexer.py Project Index · server.py MCP Server · .mcp.json Host Configuration
Components 14
Setup and Host
- Setup Orchestrator
.wavefoundry/framework/scripts/setup_wavefoundry.py - Host Configuration
.mcp.json - WF Launchers
.wavefoundry/bin/wf
Repository Intelligence
- Project Index
.wavefoundry/framework/scripts/indexer.py - Code Graph
.wavefoundry/framework/scripts/graph_indexer.py - Vector Store
.wavefoundry/framework/scripts/sqlite_vector_store.py
Engineering Workflow
- Workflow Lifecycle
.agents/skills - Project Memory
.wavefoundry/framework/scripts/memory_cli.py - Techdocs Support
.wavefoundry/framework/scripts/techdocs_audit.py
Agent Interfaces
- MCP Server
.wavefoundry/framework/scripts/server.py - MCP Tools
.wavefoundry/framework/scripts/mcp_tool_extensions.py - Workflow Dashboard
.wavefoundry/framework/dashboard/dashboard.js
Other components
- Engineer
- Coding Agent
Connections 10
- Engineer → Coding Agent: requests setup
README.md - Coding Agent → Setup Orchestrator: runs setup
README.md - Setup Orchestrator → Project Index: builds index
README.md - Setup Orchestrator → Host Configuration: writes config
README.md - Setup Orchestrator → WF Launchers: writes launchers
README.md - Setup Orchestrator → MCP Server: dry-runs server
README.md - Coding Agent → MCP Server: uses tools
- Project Index → Code Graph: indexes code
- Project Index → Vector Store: stores embeddings
- MCP Server → MCP Tools: dispatches tools