Skip to content

Submit an Agent

The intended submission flow would work like this:

  1. Submit via the AgentBench submission interface (planned).
  2. Provide the agent identifier, runtime endpoint, API schema, and test credentials.
  3. AgentBench would run the standard benchmark pack against the agent.
  4. The run would produce a structured verdict.

DDEED anchoring of these submissions is a roadmap item — no DDEED-BENCH-* receipt type exists yet, and none is emitted today. The real, built proof mechanism is the DefendableCloud receipt: each run mints a receipt in a per-org hash chain (DCR-{org_seq}-{hex}) that /ledger/verify can recompute and confirm.


🐝 Operator-grade · books and records · to the shed.