Thirdfy Agent CLI
Run Thirdfy actions from your terminal, scripts and CI with thirdfy-agent. Sign in by email, find actions, run them, and read JSON results.
thirdfy-agent lets you run Thirdfy actions from your terminal, a script, a cron job or CI. Add --json to any command and it prints a stable JSON envelope your code can read.
These docs match CLI 0.5.1.
Your first session
Step 2 does the setup for you. It creates your owner session, an agent API key and your execution wallets (the wallets your agent acts from). It saves them in ~/.thirdfy/config.json.
New profiles use runMode=agent_wallet by default, so you can run actions right away.
Step 4 returns the reading inside the JSON envelope. Here is a trimmed response from a real run on 2026-10-08:
Before your first trade
A trade moves real funds, so check the venue first.
- Check the venue. Run
thirdfy-agent venue readiness hyperliquid --json. It lists anything that blocks you (blockingReasons) and the next step. - Clear the blockers. Usually this means funding the venue.
thirdfy-agent venue fund hyperliquid --jsonreturns a funding plan. Nothing moves until you send each step as a write withrun. The one-timevenue setupruns by itself with your first order. See the venue's integration page. - Preview, then run.
thirdfy-agent previewtells you whether Thirdfy would sign the write, without sending it. Then run it: Run & status.
Chains
Run thirdfy-agent chains --json to list the chains Thirdfy acts on, with their chain ids and execution lanes (how Thirdfy runs actions on each chain).
Most actions use Base when you pass no --chain-id. Solana (chain 101) works only with agent_wallet, on your agent's managed Solana wallet. See Solana.
Guides
| Topic | Page |
|---|---|
| Install, upgrade and CI | Install |
| Email sign-in, keys and execution-only mode | Authentication |
| Find actions and check a write before you run it | Actions & preflight |
| Run reads and writes, poll status | Run & status |
| Delegation and JSON for automation | Delegation & machine output |
| Run it inside Hermes or OpenClaw | Hermes and OpenClaw |
| Every command and flag | CLI command reference |
For coding agents: llms.txt ยท llms-full.txt ยท thirdfy-agent help onboarding --json
CLI or MCP
Use the CLI in shells, scripts and CI. If your agent runs in Claude, Claude Code, Cursor, OpenAI or Grok, connect Thirdfy MCP instead.
Both reach the same action catalog and run the same checks. The same agent key works in both.