Files
stack/docs/scratchpads/85-connect-disconnect-protocol.md
Jason Woltje 12abdfe81d feat(#93): implement agent spawn via federation
Implements FED-010: Agent Spawn via Federation feature that enables
spawning and managing Claude agents on remote federated Mosaic Stack
instances via COMMAND message type.

Features:
- Federation agent command types (spawn, status, kill)
- FederationAgentService for handling agent operations
- Integration with orchestrator's agent spawner/lifecycle services
- API endpoints for spawning, querying status, and killing agents
- Full command routing through federation COMMAND infrastructure
- Comprehensive test coverage (12/12 tests passing)

Architecture:
- Hub → Spoke: Spawn agents on remote instances
- Command flow: FederationController → FederationAgentService →
  CommandService → Remote Orchestrator
- Response handling: Remote orchestrator returns agent status/results
- Security: Connection validation, signature verification

Files created:
- apps/api/src/federation/types/federation-agent.types.ts
- apps/api/src/federation/federation-agent.service.ts
- apps/api/src/federation/federation-agent.service.spec.ts

Files modified:
- apps/api/src/federation/command.service.ts (agent command routing)
- apps/api/src/federation/federation.controller.ts (agent endpoints)
- apps/api/src/federation/federation.module.ts (service registration)
- apps/orchestrator/src/api/agents/agents.controller.ts (status endpoint)
- apps/orchestrator/src/api/agents/agents.module.ts (lifecycle integration)

Testing:
- 12/12 tests passing for FederationAgentService
- All command service tests passing
- TypeScript compilation successful
- Linting passed

Refs #93

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
2026-02-03 14:37:06 -06:00

226 lines
8.4 KiB
Markdown

# Issue #85: [FED-002] CONNECT/DISCONNECT Protocol
## Objective
Implement the connection handshake protocol for federation, building on the Instance Identity Model from issue #84. This includes:
- Connection request/accept/reject handshake
- Message signing and verification using instance keypairs
- Connection state management (PENDING → ACTIVE, DISCONNECTED)
- API endpoints for initiating and managing connections
- Proper error handling and validation
## Context
Issue #84 provides the foundation:
- `Instance` model with keypair for signing
- `FederationConnection` model with status enum (PENDING, ACTIVE, SUSPENDED, DISCONNECTED)
- `FederationService` with identity management
- `CryptoService` for encryption/decryption
- Database schema is already in place
## Approach
### 1. Create Types for Connection Protocol
Define TypeScript interfaces in `/apps/api/src/federation/types/connection.types.ts`:
```typescript
// Connection request payload
interface ConnectionRequest {
instanceId: string;
instanceUrl: string;
publicKey: string;
capabilities: FederationCapabilities;
timestamp: number;
signature: string; // Sign entire payload with private key
}
// Connection response
interface ConnectionResponse {
accepted: boolean;
instanceId: string;
publicKey: string;
capabilities: FederationCapabilities;
reason?: string; // If rejected
timestamp: number;
signature: string;
}
// Disconnect request
interface DisconnectRequest {
instanceId: string;
reason?: string;
timestamp: number;
signature: string;
}
```
### 2. Add Signature Service
Create `/apps/api/src/federation/signature.service.ts` for message signing:
- `sign(message: object, privateKey: string): string` - Sign a message
- `verify(message: object, signature: string, publicKey: string): boolean` - Verify signature
- `signConnectionRequest(...)` - Sign connection request
- `verifyConnectionRequest(...)` - Verify connection request
### 3. Create Connection Service
Create `/apps/api/src/federation/connection.service.ts`:
- `initiateConnection(workspaceId, remoteUrl)` - Start connection handshake
- `acceptConnection(workspaceId, connectionId)` - Accept pending connection
- `rejectConnection(workspaceId, connectionId, reason)` - Reject connection
- `disconnect(workspaceId, connectionId, reason)` - Disconnect active connection
- `getConnections(workspaceId, status?)` - List connections
- `getConnection(workspaceId, connectionId)` - Get single connection
### 4. Add API Endpoints
Extend `FederationController` with:
- `POST /api/v1/federation/connections/initiate` - Initiate connection to remote instance
- `POST /api/v1/federation/connections/:id/accept` - Accept incoming connection
- `POST /api/v1/federation/connections/:id/reject` - Reject incoming connection
- `POST /api/v1/federation/connections/:id/disconnect` - Disconnect active connection
- `GET /api/v1/federation/connections` - List workspace connections
- `GET /api/v1/federation/connections/:id` - Get connection details
- `POST /api/v1/federation/incoming/connect` - Public endpoint for receiving connection requests
### 5. Connection Handshake Flow
**Initiator (Instance A) → Target (Instance B)**
1. Instance A calls `POST /api/v1/federation/connections/initiate` with `remoteUrl`
2. Service creates connection record with status=PENDING
3. Service fetches remote instance identity from `GET {remoteUrl}/api/v1/federation/instance`
4. Service creates signed ConnectionRequest
5. Service sends request to `POST {remoteUrl}/api/v1/federation/incoming/connect`
6. Instance B receives request, validates signature
7. Instance B creates connection record with status=PENDING
8. Instance B can accept (status=ACTIVE) or reject (status=DISCONNECTED)
9. Instance B sends signed ConnectionResponse back to Instance A
10. Instance A updates connection status based on response
### 6. Security Considerations
- All connection requests must be signed with instance private key
- All responses must be verified using remote instance public key
- Timestamps must be within 5 minutes to prevent replay attacks
- Connection requests must come from authenticated workspace members
- Public key must match the one fetched from remote instance identity endpoint
### 7. Testing Strategy
**Unit Tests** (TDD approach):
- SignatureService.sign() creates valid signatures
- SignatureService.verify() validates signatures correctly
- SignatureService.verify() rejects invalid signatures
- ConnectionService.initiateConnection() creates PENDING connection
- ConnectionService.acceptConnection() updates to ACTIVE
- ConnectionService.rejectConnection() marks as DISCONNECTED
- ConnectionService.disconnect() updates active connection to DISCONNECTED
- Timestamp validation rejects old requests (>5 min)
**Integration Tests**:
- POST /connections/initiate creates connection and calls remote
- POST /incoming/connect validates signature and creates connection
- POST /connections/:id/accept updates status correctly
- POST /connections/:id/reject marks connection as rejected
- POST /connections/:id/disconnect disconnects active connection
- GET /connections returns workspace connections
- Workspace isolation (can't access other workspace connections)
## Progress
- [x] Create scratchpad
- [x] Create connection.types.ts with protocol types
- [x] Write tests for SignatureService (18 tests)
- [x] Implement SignatureService (sign, verify, validateTimestamp)
- [x] Write tests for ConnectionService (20 tests)
- [x] Implement ConnectionService (all 8 methods)
- [x] Write tests for connection API endpoints (13 tests)
- [x] Implement connection API endpoints (7 endpoints)
- [x] Update FederationModule with new providers
- [x] Verify all tests pass (70/70 passing)
- [x] Verify type checking passes
- [x] Verify test coverage ≥85% (100% coverage on new code)
- [x] Commit changes (commit fc39190)
## Testing Plan
### Unit Tests
1. **SignatureService**:
- Should create RSA SHA-256 signatures
- Should verify valid signatures
- Should reject invalid signatures
- Should reject tampered messages
- Should reject expired timestamps
2. **ConnectionService**:
- Should initiate connection with PENDING status
- Should fetch remote instance identity before connecting
- Should create signed connection request
- Should accept connection and update to ACTIVE
- Should reject connection with reason
- Should disconnect active connection
- Should list connections for workspace
- Should enforce workspace isolation
### Integration Tests
1. **POST /api/v1/federation/connections/initiate**:
- Should require authentication
- Should create connection record
- Should fetch remote instance identity
- Should return connection details
2. **POST /api/v1/federation/incoming/connect**:
- Should validate connection request signature
- Should reject requests with invalid signatures
- Should reject requests with old timestamps
- Should create pending connection
3. **POST /api/v1/federation/connections/:id/accept**:
- Should require authentication
- Should update connection to ACTIVE
- Should set connectedAt timestamp
- Should enforce workspace ownership
4. **POST /api/v1/federation/connections/:id/reject**:
- Should require authentication
- Should update connection to DISCONNECTED
- Should store rejection reason
5. **POST /api/v1/federation/connections/:id/disconnect**:
- Should require authentication
- Should disconnect active connection
- Should set disconnectedAt timestamp
6. **GET /api/v1/federation/connections**:
- Should list workspace connections
- Should filter by status if provided
- Should enforce workspace isolation
## Design Decisions
1. **RSA Signatures**: Use RSA SHA-256 for signing (matches existing keypair format)
2. **Timestamp Validation**: 5-minute window to prevent replay attacks
3. **Workspace Scoping**: All connections belong to a workspace for RLS
4. **Stateless Protocol**: Each request is independently signed and verified
5. **Public Connection Endpoint**: `/incoming/connect` is public (no auth) but requires valid signature
6. **State Transitions**: PENDING → ACTIVE, PENDING → DISCONNECTED, ACTIVE → DISCONNECTED
## Notes
- Connection requests are workspace-scoped (authenticated users only)
- Incoming connection endpoint is public but cryptographically verified
- Need to handle network errors gracefully when calling remote instances
- Should validate remote instance URL format before attempting connection
- Consider rate limiting for incoming connection requests (future enhancement)