How saiakashneela/projectmesh works
Projectmesh is a local AI engineering platform for a repository-scoped Architect-to-Executor handoff. An AI client inspects a selected workspace and records tasks and project context through MCP; Projectmesh packages the active task and affected source for a locally approved executor, then records execution state and a review. The CLI also initializes and analyzes repositories, while a multi-workspace registry and optional HTTP/ngrok sharing support concurrent remote access. The graph follows documented capabilities where source excerpts are incomplete; uncertain internal wiring is omitted.
Main entry points: cli.ts Projectmesh CLI · ai-workspace.ts AI Workspace Flows · share.ts HTTP Sharing · mcp-tools.ts Platform Tools
Components 20
User Access
- Projectmesh CLI
src/cli.ts - MCP Server
src/mcp-server.ts - HTTP Sharing
src/share.ts - Dashboard
src/dashboard.ts - Platform Tools
src/mcp-tools.ts - ngrok Tunnel
Planning Handoff
- Repository Analysis
src/repository-analysis.ts - AI Workspace Flows
src/ai-workspace.ts - Task Context
- Task Packet
Execution Review
- Agent Runners
src/agent-runners.ts - Execution Records
Workspace Platform
- Workspace Access
src/workspace.ts - Workspace Registry
src/platform-config.ts - Git Operations
src/git.ts - Target Repository
- Platform State
Other components
- Architect AI
- Local User
- Executor Agent
Connections 31
- Local User → Projectmesh CLI: runs commands
- Architect AI → MCP Server: connects tools
- MCP Server → Platform Tools: dispatches tools
- MCP Server → Workspace Registry: resolves workspaces
- Platform Tools → Repository Analysis: requests inspection
- Platform Tools → AI Workspace Flows: manages handoff
- Platform Tools → Git Operations: runs git tools
- Repository Analysis → Workspace Access: reads workspace
- Repository Analysis → Target Repository: inspects files
- AI Workspace Flows → Workspace Access: reads and writes
- AI Workspace Flows → Task Context: updates tasks
- AI Workspace Flows → Task Packet: generates packet
- Projectmesh CLI → Repository Analysis: analyzes repository
- Projectmesh CLI → AI Workspace Flows: runs handoff flows
- Projectmesh CLI → Workspace Registry: selects workspace
- Projectmesh CLI → HTTP Sharing: starts services
- Projectmesh CLI → Agent Runners: selects executor
- Projectmesh CLI → Git Operations: captures changes
- Local User → Projectmesh CLI: approves execution
- Projectmesh CLI → Execution Records: records execution
- Agent Runners → Executor Agent: launches agent
- Executor Agent → Target Repository: changes files
- Projectmesh CLI → AI Workspace Flows: writes report
- HTTP Sharing → MCP Server: serves HTTP MCP
- HTTP Sharing → Dashboard: starts dashboard
- HTTP Sharing → ngrok Tunnel: starts tunnel
- Dashboard → Workspace Registry: reads workspace state
- Dashboard → ngrok Tunnel: configures token
- Workspace Registry → Platform State: reads and writes
- Git Operations → Target Repository: reads git state
- Workspace Access → Target Repository: confines file access