| name | artifact-registry |
| description | Use this skill when Prometheus needs to find, index, normalize, retrieve, or reason about generated artifacts from chats, tool observations, background tasks, teams, Brain runs, processes, media generations, canvas exports, or Creative outputs. Triggers on phrases like find artifact, artifact registry, generated outputs, where did that file go, task artifacts, team artifacts, index outputs, artifact search, canvas export, and generated media history. Use it to map scattered Prometheus output sources into durable searchable artifact records. |
| emoji | 🗂️ |
| version | 1.1.0 |
| triggers | find artifact, artifact registry, generated outputs, where did that file go, task artifacts, team artifacts, index outputs, artifact search, canvas export, generated media history, output registry, artifact index |
Artifact Registry
Use this skill when Prometheus needs to find, index, or reason about generated outputs.
Current State
Status: partial. Artifacts exist in several places, but there is no single first-class registry/query tool yet.
Current sources:
- Chat messages:
artifacts, generatedImages, generatedVideos, processEntries in src/gateway/session.ts
- Tool observations:
ToolObservation.artifacts, pathsTouched, resultRawRef
- Task evidence bus: category
artifact
- Team artifacts:
share_artifact and team room state
- Legacy DB table:
src/db/database.ts has artifacts, but current chat/runtime artifacts are not all unified there
- Brain artifacts:
workspace/Brain/...
- Canvas/Creative export artifacts in
src/gateway/routes/canvas.router.ts
Procedure
- Search the active session first.
- Read tool observations for artifact arrays and path references.
- If task-bound, load task evidence bus and journal.
- If team-bound, inspect team shared artifacts.
- If creative/media, inspect canvas/creative export metadata.
- Write a normalized artifact index entry with path/url, source session/task/team, type, createdAt, and verification status.
Gap To Implement
Add a durable artifact_index with list/search/read APIs and a background indexer over sessions, tool observations, tasks, teams, and Brain outputs.
