validate

To record validation results, you create a manifest capturing the run's id, candidate_set_id, work_item_ids, candidate_branch, candidate_commit, timing, status, and checks, with optional evidence references attached to formalize the outcome.

A validation run manifest is a record that captures a validation run's results and links them to relevant work items. You need to create one when validators finish checks or when the orchestrator needs to record results for candidate sets.

You update work item status by applying the validation outcome to drive orchestration transitions. This requires a two-phase write workflow that writes the manifest first and then updates the orchestration state via state-commit.

Environment_error is handled as a non-code failure path, distinguishing infrastructure or environment issues from a failed status which indicates the validation checks themselves did not pass for the candidate set.

Recording validation outcomes requires an initialized orchestration, an active candidate set, and a two-phase write workflow that writes the manifest first and then updates the orchestration state via state-commit, with optional evidence verification.

Yes, evidence verification is optional during the two-phase write workflow. You write the validation manifest first, verify the attached evidence references, and then update the orchestration state via state-commit.