| name | ref-setup |
| description | Download AI reference materials (external source code for AI assistant reference). Offers individual or all references. |
| allowed-tools | Bash(bash ref/*), Bash(ls *), Bash(test *), Bash(tail *), Bash(sleep *), Read, AskUserQuestion |
Step 1 โ Ask which reference to install
Use AskUserQuestion with the following options:
| Option | Label | Description |
|---|
| 1 | all | Download all reference materials (default) |
| 2 | datahub | Download only DataHub source code |
| 3 | clean | Remove all downloaded content |
Map the selected option to a ref/setup.sh argument:
- all โ
bash ref/setup.sh (no argument)
- datahub โ
bash ref/setup.sh datahub
- clean โ
bash ref/setup.sh --clean
Step 2 โ Run setup
Execute the setup script in the background: bash ref/setup.sh <selected-option>
- Note the background task ID and output file path.
Step 3 โ Monitor progress
While the script runs, check the background task output every ~20 seconds:
- Read the output file (e.g.,
tail -30 <output-file>) to report progress messages (cloning, cleanup steps, etc.).
- After each check, summarize what has happened since the last check.
- If an error is printed or the script exits with a non-zero code, report it immediately with the full output and stop.
Continue until the background script exits with exit code 0.
Step 4 โ Report to user
When the script exits successfully:
- Tell the user that all reference materials are ready.
- Show the location:
ref/github/datahub/
- Remind the user of the key directories available for AI reference:
- DataHub (
ref/github/datahub/):
metadata-models/ โ Entity schemas (PDL/Avro)
metadata-service/ โ GMS backend (Java/Spring)
datahub-web-react/ โ Frontend (TypeScript/React)
metadata-ingestion/ โ Python SDK and ingestion framework
datahub-graphql-core/ โ GraphQL API schemas