What is EarnClaw
EarnClaw is the workspace to create, deploy, fund, and monitor finance agents — with validated trading, yield, and prediction templates, hosted runtimes, and market data integrations.
EarnClaw is where you ship finance agents: pick a template, choose a runtime (Hermes, OpenClaw, or Claude), set guardrails, and operate from one dashboard.
You get certified strategy packs, org policies, wallet delegation, run history, and billing in one place. Every venue and data integration your agent uses routes through the Thirdfy action catalog, called from hosted runtimes via the Thirdfy Agent CLI or Thirdfy MCP. You do not rebuild execution rails or wire providers one by one.
We believe autonomous agents will reshape how finance runs: execution, risk, and operations move from manual clicks to governed, repeatable cycles. EarnClaw is built for that shift with production-grade deployment frameworks, hosted runtimes, policy guardrails, and LLM decision layers that fail closed when preflight does not pass.
How EarnClaw is built
Professional agents in three lanes
EarnClaw ships validated agents in the lanes finance teams actually run today. Trading, yield, and prediction venues all use the same Thirdfy catalog surface (Agent CLI on Hermes and OpenClaw, MCP on Claude managed). Prediction today means hosted Polymarket signal agents on Hermes. Venue setup, funding, and CLOB readiness are covered in Thirdfy Polymarket integration.
Integrations and context
Agents need fresh context before they decide, and governed paths when they act. All integrations (market data, trading venues, yield providers, and Polymarket) are catalog actions on Thirdfy. Hosted runtimes reach them through the Thirdfy Agent CLI or Thirdfy MCP, not separate per-provider SDKs in EarnClaw.
Detail: Integrations at runtime. Catalog reference: Thirdfy integrations hub.
Market signals and prices
Yield intelligence
Trading venues
Polymarket
Who it is for
Finance and crypto users
Builders
Operators
Partners
What happens at runtime
On each cycle, the runtime reads live evidence, applies template policy, and writes an execute-or-skip decision. Delegation and guardrails limit what the agent may attempt. Before any onchain action runs, policy preflight must pass.
For integrations, preflight, and runtime behavior, see Integrations at runtime. You do not need integrator docs to create your first agent.