claude-flow
Version:
Ruflo - Enterprise AI agent orchestration for Claude Code. Deploy 60+ specialized agents in coordinated swarms with self-learning, fault-tolerant consensus, vector memory, and MCP integration
21 lines • 1.02 kB
TypeScript
/**
* GAIA DAG Harness — Smoke Tests (ADR-139 Addendum)
*
* Verifies the Co-Sight DAG architecture without live API calls ($0 cost).
*
* Test cases:
* 1. DAG parse — valid JSON → DagPlan with correct step structure
* 2. DAG parse — malformed JSON fallback → single-step plan
* 3. getReadySteps — step 0 ready (no deps), step 1 NOT ready (dep=0 not done)
* 4. getReadySteps — after step 0 completed, step 1 becomes ready
* 5. getReadySteps — parallel steps (0 and 1 both no deps → both ready)
* 6. Parallel execution — all ready steps fire concurrently (mock)
* 7. Blocked-step detection — step with blocked dep is NOT ready
* 8. Plan cap — more than 7 steps is capped to MAX_PLAN_STEPS=7
* 9. runGaiaDAG mock — full mock pipeline returns expected answer
* 10. Finalizer extraction — FINAL_ANSWER in finalize response is extracted
*
* Refs: ADR-139, github.com/ZTE-AICloud/Co-Sight, #2156
*/
export {};
//# sourceMappingURL=gaia-dag.smoke.d.ts.map