Dispatch a subagent to adversarially audit a PR (or the current diff) for risks, regressions, assumptions, gaps, bugs, issues, behaviour changes, leaks, and second-order consequences. Use before merging.
Write a canonical session-handoff doc and a copy-paste kickoff message so work resumes cleanly in a new session. Use at end of session, before a context reset, or when told to write the handoff.
Record what a session touched in the analyze-service index store, and write or validate an entry. The protocol /handoff follows at end of session; rarely run directly.
Report which shipped productivity tools are actually USED and YIELDING versus idle or DEAD. Use for "is anyone using X", "which tools are dead", "did the tool catch anything", or whether a shipped tool stuck.
Operate auditloop — the hosted UX-audit crawler at auditloop.zacx.dev (multi-viewport crawl, axe a11y, perf/web-vitals, run diffs and trends, vision-LLM notes, login recipes). Deploy it, manage targets/runs/plugin tokens, read the cost dashboard. Use for:…
Operate the homelab Tekton CI/CD platform — Pipelines/Triggers/Dashboard, the GitHub webhook (tekton-webhook.zacx.dev), the naida-ux-audit pipeline, and clawgate-ci (the real pre-merge gate). Use for: Tekton, the homelab CI platform, clawgate-ci, the Tekton…
Run the re-runnable naida + vetr UX-audit harnesses — walk the app's funnel, screenshot every view, capture console/network/axe findings, then re-run to verify. Local-only. Use for: UX-audit or QA-sweep naida or vetr, "make ux-audit", the funnel review, the…
Deterministic post-agent verification gate: re-run the AUTHORITATIVE build/typecheck/test/vet gate + git-completeness + stale-deps checks against a repo/worktree, and READ that verdict instead of trusting an agent's 'done' claim.