Quickstart
Email OTP onboarding via Agent CLI or MCP. Discover the catalog, preflight, and run your first intent.
Architecture: See Architecture for the flow before diving in.
Email onboarding (CLI or MCP)
For AI agents: Use llms.txt for docs discovery, or llms-full.txt for the full API spec in one file (Execute Intent + x402). Read one first.
1. Email OTP login
CLI:
MCP: startEmailOnboarding → completeEmailOnboarding (acceptTerms: true).
2. Confirm credentials
CLI: thirdfy-agent whoami --json (stores agentApiKey, owner session, executionWallets).
MCP: Response may include agent credentials; then getWalletProfile.
3. Discover actions
CLI: thirdfy-agent actions --json
MCP: getActionsCatalog or getProviderActions
Public catalog (no auth):
4. Preflight
CLI: thirdfy-agent preflight --run-mode agent_wallet ...
MCP: preflight tools before your first write.
5. Execute
CLI: thirdfy-agent run --run-mode agent_wallet ...
MCP: walletExecute or agentRun
Fund the execution wallet for your venue before live writes.
6. Poll status
CLI: thirdfy-agent intent-status --intent-id <id> --json
MCP: getIntentStatus
HTTP:
Details: Quick start · CLI authentication · MCP onboarding
Full Example: Success and Rejection
Success response (HTTP 200):
Poll GET /api/v1/agent/execute-intent/status?intentId=int_abc123 until status is completed or failed.
Rejection response (HTTP 4xx):
Common rejection reasons: invalid API key, action not in allowlist, params mismatch, estimatedAmountUsd exceeds policy limit. See Troubleshooting for remediation.
Environment: Set THIRDFY_API_URL=https://api.thirdfy.com (production) or https://api-test.thirdfy.com (testnet). Export THIRDFY_AGENT_API_KEY from email onboarding (whoami or MCP response). Credits: Delegated users pay for execution. See Credits for Users.
Optional: Creator Platform
Use thirdfy.com/creator later for directory profile, skills package upload, and production review. See Creator Platform Guide.
Troubleshooting
Reference
- llms.txt / llms-full.txt: llms.txt = docs discovery. llms-full.txt = full API spec in one file.
- Troubleshooting: Common errors and fixes
- FAQ: Common questions
Next Steps
- Agent CLI or MCP: Primary integration surfaces
- Install Skills: Optional OpenClaw skill manifest
- Integrating with Execute Intents: Validation, fan-out, security