Handbook
Doc-hydration docs-check job (Hydration v2)
Fleet is the template-only execution rail for documentation quality runs in the Forge Hydration v2 pipeline: site build, link check, and scorecard jobs run as docker_argv jobs against an approved requirement template —…
Updated
Files
| File | Purpose |
|---|---|
requirement-template.json |
Requirement template row set to install via PUT /v1/container-templates (or admin UI): docs_check_worker, pinned Python image. |
job-docs-check-request.json |
POST /v1/jobs payload: workspace-uploaded docs repo, template image injection, then build + link check + scorecard inside the container. |
Flow
- Install the requirement template once:
curl -sS -X PUT "$FLEET/v1/container-templates" \
-H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
--data @requirement-template.json
- Submit the job (uses
meta.use_fleet_template_image+meta.requirements, withworkspace_upload_requiredso the docs repo snapshot is uploaded viaPUT /v1/jobs/{id}/workspace— seedocs/build-201/01-workspace-upload.md):
curl -sS -X POST "$FLEET/v1/jobs" \
-H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
--data @job-docs-check-request.json
- Job stdout/artifacts include the build log,
link_report.json, and the scorecard output — these attach to the ForgeRun evidence for the hydration cycle (forge-platform/scripts/docs_scorecard.pyconsumes them).
Governance
- Template-only rail: the job image comes from the approved requirement template; the argv only selects which documented check scripts to run.
- Results are evidence, not decisions: promotion still requires the reviewer decision manifest in forge-platform.