Best for: Building agents or automation against Traseq
Drive Traseq from an agent or API client
The same authoring contract — SignalGraph v2 — that the no-code editor produces is what the API expects. Your agent and the UI cannot drift.
Key concepts
- API key
- Generated in your workspace settings. Each plan caps how many keys a single workspace can hold at once.
- Capabilities
- Read /capabilities first. It tells your agent which fields, operators, and instruments your workspace supports right now — no hardcoding.
- SignalGraph validation
- Every strategy write is checked against schema and semantics. Invalid graphs are rejected with a structured diff so your agent can self-repair.
- Create / finalize
- Authoring is two-step: create a draft, then finalize it to lock a versioned snapshot. Backtests only run against finalized versions.
- Queue backtest
- Backtests run async on the engine. Submit, get a run id, then poll the run or stream the result event.
How this looks in Traseq
Generate an API key in your workspace settings.
Call /capabilities to confirm your environment matches what your agent expects.
Build a SignalGraph v2 strategy, validate it, then finalize a version.
Queue a backtest against that finalized version, then read the run.
Your next step