← ClaudeAtlas

dashboard-snapshotlisted

Assemble a dashboard-ready snapshot from the canonical state.
panjose/Co-Scientist · ★ 4 · AI & Automation · score 74
Install: claude install-skill panjose/Co-Scientist
# dashboard-snapshot Goal: - Assemble a dashboard-ready snapshot from the canonical state. Inputs: - `CoScientistState` - `run_id` - `updated_at` Outputs: - `dashboard/SNAPSHOT.json` Context Loading: - Read `skills/shared-references/artifact-contract.md` so the dashboard artifact stays downstream of canonical run artifacts. - Read `packages/dashboard_contracts/snapshot_projection.py` before rebuilding `dashboard/SNAPSHOT.json`. - Read `packages/dashboard_contracts/dashboard.py` when the dashboard-facing snapshot field layout needs to be checked. Execution Contract: - This skill is deterministic and must not call an LLM. - The canonical shared builder surface is `from tools import build_dashboard_snapshot`. - The underlying implementation lives in `packages/dashboard_contracts/snapshot_projection.py`. - The snapshot builder is a pure projection shared by API reads and artifact writes. - The dashboard layer must remain downstream of canonical state and artifact contracts and must not invent new pipeline semantics. Execution Steps: 1. Open `skills/shared-references/artifact-contract.md`, then read `packages/dashboard_contracts/snapshot_projection.py`. 2. Load the canonical run state and routing artifacts required to build the dashboard projection. 3. Call the canonical snapshot builder and write `dashboard/SNAPSHOT.json`. 4. Keep the emitted snapshot aligned with the dashboard-facing contract models. Completion Rule: - This skill is complete only when `dashboard/SN