These are the end-to-end outputs the assistant produces when an operator brings a task, one per deliverable
mode shown in ../DEMO.md. Think of them as illustrative artifacts, what the skills generate rather
than new knowledge, useful for (a) checking the output format and (b) onboarding.
All three assume the active profile is example-site (DC_SITE_PROFILE=example-site,
see ../skills/site-profile/profiles/example-site.yaml) and a base already positioned at the work site
(per scope). Each shows the full chain: request → clarify (resolve from profile, else ask) →
feasibility → deliverable → verification → abort/escalation.
| Mode | File | Owning skill | Deliverable |
|---|---|---|---|
| Task plan | task-plan-patch-cable-connect.md |
datacenter-task-planner (+ connector-mate / cable-route patterns) |
feasibility + manipulation plan |
| SOP run | sop-cable-cleanup.md |
sop-execution (+ sop-cable-cleanup pattern) |
reconcile-to-baseline + audit |
| Inspection assessment | inspection-thermal-switch-triage.md |
inspection-judgment (+ thermal-switch-triage pattern) |
per-target assessment + escalation |
They trace the same RJ45 patch-panel rack as the demo tasks (../DEMO.md). The numbers stay consistent with
openarm-embodiment limits, and site values come from the example-site profile.