How pradhan-not-found/checkpost works
Checkpost is an AI-agent security and monitoring dashboard. Its principal documented flow sends an agent request through the proxy, evaluates policy and usage controls, then forwards permitted requests to an LLM provider and returns the response; the proxy also records usage and traces. Operators manage agents, policies, monitoring, settings, and held actions through the dashboard. The proxy and Firestore helpers are evidenced in source excerpts; several dashboard and HITL relationships are documented in the README, while their runtime implementation was not sampled.
Main entry points: db.ts Firestore state · route.ts Request proxy · DatabaseContext.tsx Live database context · dashboard Dashboard
Components 22
Operator dashboard
- Dashboard
src/app/dashboard - Agent management
src/app/dashboard/agents/page.tsx - System overview
src/app/dashboard/page.tsx - Audit logs
src/app/dashboard/logs/page.tsx - Policy management
src/app/dashboard/policies/page.tsx - Provider settings
src/app/dashboard/settings/page.tsx - LLM testing
src/app/dashboard/llm-test/page.tsx - Threat monitoring
src/app/dashboard/threats/page.tsx
Agent control plane
- Request proxy
src/app/api/proxy/route.ts - Policy and spend checks
src/app/api/proxy/route.ts - Human review queue
src/app/chapter/hitl/page.tsx - Agent registration API
src/app/api/agents/route.ts - Agent policy API
src/app/api/agent/policies/route.ts - Trace ingestion API
src/app/api/agent/ingest/route.ts - Dashboard data API
src/app/api/db/route.ts
State and identity
- Firestore state
src/lib/db.ts - Live database context
src/context/DatabaseContext.tsx - Authentication
src/context/AuthContext.tsx - Firebase clients
src/lib/firebase.ts
Other components
- Admin operator
- AI agent
- External LLM providers
Connections 27
- Admin operator → Dashboard: uses
README.md - AI agent → Request proxy: submits request
README.md - Request proxy → Policy and spend checks: evaluates request
README.md - Policy and spend checks → Firestore state: loads rules
README.md - Policy and spend checks → Human review queue: holds sensitive action
README.md - Admin operator → Human review queue: approves or rejects
README.md - Policy and spend checks → External LLM providers: forwards allowed request
README.md - Request proxy → External LLM providers: calls provider
src/app/api/proxy/route.ts - External LLM providers → Request proxy: returns output
README.md - Request proxy → AI agent: returns response
README.md - Request proxy → Firestore state: reads and records usage
src/app/api/proxy/route.ts - Agent registration API → Firestore state: saves agents
src/app/api/agents/route.ts - Agent policy API → Firestore state: uses database
src/app/api/agent/policies/route.ts - Trace ingestion API → Firestore state: uses database
src/app/api/agent/ingest/route.ts - Dashboard data API → Firestore state: reads and updates
src/app/api/db/route.ts - Firestore state → Firebase clients: persists through
src/lib/db.ts - Live database context → Firebase clients: subscribes to Firestore
src/context/DatabaseContext.tsx - Dashboard → Live database context: syncs state
README.md - Dashboard → Authentication: uses authentication
README.md - Agent management → Live database context: reads state
src/app/dashboard/agents/page.tsx - System overview → Live database context: reads state
src/app/dashboard/page.tsx - Audit logs → Live database context: reads state
src/app/dashboard/logs/page.tsx - Policy management → Live database context: reads state
src/app/dashboard/policies/page.tsx - Provider settings → Authentication: uses authentication
src/app/dashboard/settings/page.tsx - Admin operator → Agent management: manages agents
README.md - Admin operator → Policy management: configures policies
README.md - Admin operator → Audit logs: reviews traces
README.md