Persist Ward skill-use artifacts as trajectory events #54

New issue

Closed

opened 2026-07-24 18:54:18 +00:00 by coilyco-ops · 0 comments

coilyco-ops commented

2026-07-24 18:54:18 +00:00

Member

Goal

Complete the documented app/skill_use.py migration by retaining normalized Ward reap skill-use evidence in the contract-v1 append-only trajectory store while preserving the current structured-log and Prometheus projections.

Acceptance criteria

Map each normalized skill-use record to a metadata-only observation.recorded trajectory event.
Preserve Ward run, request, repository, issue, workflow, role, and harness joins when present.
Make repeated artifact ingestion idempotent in the durable store.
Keep skill bodies, local paths, container names, and other source content out of retained envelopes.
Wire the existing startup artifact importer to the configured trajectory database.
Add focused tests and update the shipped feature and migration inventories.

## Goal Complete the documented `app/skill_use.py` migration by retaining normalized Ward reap skill-use evidence in the contract-v1 append-only trajectory store while preserving the current structured-log and Prometheus projections. ## Acceptance criteria * Map each normalized skill-use record to a metadata-only `observation.recorded` trajectory event. * Preserve Ward run, request, repository, issue, workflow, role, and harness joins when present. * Make repeated artifact ingestion idempotent in the durable store. * Keep skill bodies, local paths, container names, and other source content out of retained envelopes. * Wire the existing startup artifact importer to the configured trajectory database. * Add focused tests and update the shipped feature and migration inventories.

coilyco-ops referenced this issue from a commit

2026-07-24 18:56:25 +00:00

feat: retain Ward skill-use trajectories (closes #54)

coilyco-ops closed this issue

2026-07-24 18:56:25 +00:00

coilyco-ops referenced this issue

2026-07-24 19:07:01 +00:00

[interactive] Deploy trajectory ledger, evaluate LiteLLM, and verify first activation #56