How bronz3beard/ai.tech-lead-stack works
The Lead Stack combines an agent-agnostic workflow/tooling product with a web dashboard for repository work, AI assistance, design review, and workflow analytics. A principal web workflow is a developer submitting a brief to the Reflexion endpoint, which configures a model runner, executes iterative planning, persists run state, and returns a handoff or interview result. The README also describes intent-to-release skills and review/testing workflows. The tree and sampled files support several dashboard and MCP paths; skill catalogue internals and some integrations were not sampled, so uncertain wiring is omitted.
Main entry points: route.ts Reflexion endpoint · prisma.ts Application database · engine.ts Reflexion engine · pipeline.ts Pipeline planning
Components 20
Dashboard workflows
- Dashboard UI
apps/dashboard/src/app/layout.tsx - Chat interface
apps/dashboard/src/components/chat/ChatBody.tsx - AI chat API
apps/dashboard/src/app/api/chat/route.ts - Feature discovery
apps/dashboard/src/app/api/orchestrator/discovery/route.ts - Sandbox boot
apps/dashboard/src/app/api/orchestrator/sandbox/boot/route.ts - Design review
apps/dashboard/src/app/api/design-review/route.ts - Project management
apps/dashboard/src/app/api/projects/route.ts - Workflow analytics
apps/dashboard/src/lib/agentic-metrics.ts
AI orchestration
- Reflexion endpoint
apps/dashboard/src/app/api/orchestrator/reflexion/route.ts - Reflexion engine
packages/core/src/lib/ai/reflexion/engine.ts - User model runner
packages/core/src/lib/ai/reflexion/providers-user.ts
MCP and skills
- Pipeline planning
packages/core/src/mcp-server/handlers/pipeline.ts - MCP server
packages/core/src/mcp-server/index.ts - Skill services
packages/core/src/lib/skills/fs-service.ts
State and integrations
- Application database
packages/core/src/lib/prisma.ts - GitHub
- E2B sandbox
- AI model provider
- Telemetry
packages/core/src/lib/telemetry-service.ts
Other components
- Developer
Connections 24
- Developer → Dashboard UI: uses
- Dashboard UI → Chat interface: opens chat
- Chat interface → AI chat API: sends messages
- AI chat API → AI model provider: streams responses
- Developer → Reflexion endpoint: submits brief
- Reflexion endpoint → GitHub: reads project files
- Reflexion endpoint → User model runner: configures runner
- Reflexion endpoint → Application database: creates run state
- Reflexion endpoint → Reflexion engine: runs loop
- Reflexion engine → User model runner: generates and critiques
- Reflexion engine → Application database: persists loop state
- Developer → Feature discovery: starts discovery
- Feature discovery → AI model provider: streams assistance
- Developer → Sandbox boot: boots sandbox
- Sandbox boot → E2B sandbox: runs project
- Developer → Design review: manages review
- Design review → Application database: reads and updates sessions
- Developer → Project management: manages projects
- Project management → Application database: reads and writes projects
- Developer → MCP server: requests tools
- MCP server → Pipeline planning: dispatches pipeline
- Pipeline planning → Skill services: loads skill graph
- Pipeline planning → Telemetry: tracks planning
- Developer → Workflow analytics: views metrics