Add evaluation and annotation records to the trajectory plane #45

Closed
opened 2026-07-23 05:32:29 +00:00 by coilyco-ops · 0 comments
Member

Parent: #40

Define and ingest evaluation, verifier, annotation, and human-intervention records that join to materialized trajectories without changing Ward authorization or execution authority.

Dependencies:

  • #42 trajectory schema package and validation fixtures
  • #44 episode and trajectory materialization

Acceptance criteria:

  • Model evaluator inputs, outputs, rubric or verifier version, annotator provenance, confidence, and supersession.
  • Support automatic and human records with redaction and access-tier controls.
  • Join records by stable trajectory and event identities.
  • Preserve raw evidence references and produce replayable evaluation fixtures.
  • Document how disagreement and late annotations affect derived datasets.
Parent: #40 Define and ingest evaluation, verifier, annotation, and human-intervention records that join to materialized trajectories without changing Ward authorization or execution authority. Dependencies: - #42 trajectory schema package and validation fixtures - #44 episode and trajectory materialization Acceptance criteria: - Model evaluator inputs, outputs, rubric or verifier version, annotator provenance, confidence, and supersession. - Support automatic and human records with redaction and access-tier controls. - Join records by stable trajectory and event identities. - Preserve raw evidence references and produce replayable evaluation fixtures. - Document how disagreement and late annotations affect derived datasets.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
coilyco-flight-deck/agent-proxy#45
No description provided.