| name | dataset |
| description | This skill should be used when the user types "/dataset", "/dataset list", "/dataset production", "/dataset production, priority", "/dataset clear", or "/dataset rename old-name new-name". It exposes Retriever's visible slash-command surface for dataset scoping, discovery, and renaming.
|
| metadata | {"version":"1.1.17"} |
Operates under retriever:routing. If the user's intent actually fits a different tier — another retriever:* skill, a Tier 2 slash, a Tier 3 tools.py subcommand, or (last resort) direct DB access — stop and re-route against the ladder before continuing.
Retriever /dataset
Use this skill for /dataset, /dataset list, /dataset NAME, /dataset NAME, NAME, /dataset clear, and /dataset rename OLD NEW.
Read-only fast path
For the exact read-only forms /dataset and /dataset list:
- Do not read ../search/SKILL.md.
- Do not read schema docs.
- Use the canonical plugin tool at ../tool-template/tools.py for these read-only forms.
- Run exactly one Bash command from the workspace root:
/dataset: python3 <resolved path to ../tool-template/tools.py> slash . /dataset/dataset list: python3 <resolved path to ../tool-template/tools.py> slash . /dataset list
/dataset list renders a dataset stats table with Dataset, Docs, Size, and Custodians columns.- Do not fall back to a workspace-local tool copy.
- Return stdout exactly as the entire response. No preamble. No commentary. No reformatting.
Other forms
For /dataset NAME, comma-separated dataset selection, /dataset clear, and /dataset rename OLD NEW:
- Read ../search/SKILL.md.
- Preserve comma-separated dataset selectors exactly as written; quote names with spaces when needed.
- Treat this skill as the slash command
/dataset.
- Return only the resulting Retriever state or table output. Do not add a preamble, trailing summary, or follow-up suggestion.
