| name | docs-knowledge-maintainer |
| description | Keep concise user docs and machine-readable model evidence synchronized with real behavior without losing provenance or caveats. |
| metadata | {"origin":"adapted from ZMB-UZH/omero-docker-extended at b27dbe990703d64d13e540c40cf4e122954c664d"} |
Docs Knowledge Maintainer
Use this skill whenever behavior, operating assumptions, workflows, data,
models, risk gates, UI controls, or published evidence changes.
Routing
- User scope, install, safety, and primary commands:
README.md.
- Agent tools and provenance:
docs/AGENT_WORKFLOWS.md.
- Data lineage and truth requirements:
docs/DATA_PROVENANCE_POLICY.md.
- Execution realism:
docs/LIVE_MARKET_SIMULATION.md.
- Model methodology and promotion: model research documents and generated
latest/ evidence directories.
- CLI/Windows behavior: generated command contracts plus parity tests.
- Release behavior:
docs/release.md and the manual release workflow.
Evidence Rules
- Keep the README short enough to scan; link to the nearest complete deep doc.
- Do not delete information unless it is obsolete and its replacement preserves
the operating contract.
- Generate charts from committed CSV/JSON evidence. Never edit plotted values by
hand or publish a graph without its source table and UTC span.
- Publish only the latest iteration charts plus the cumulative progress series,
while retaining truthful machine-readable provenance needed to reproduce them.
- Label unavailable, blocked, simulated, testnet, and not-run evidence plainly.
- Update docs, tests, and generated badge metadata together when a contract
changes.
Verify links, generated badge drift, relevant doc contracts, and the focused
tests for the changed behavior before publication.