theogyeezy

docs-audit

Repo-wide documentation accuracy audit — find and fix drift between living docs (README/CLAUDE.md/checklists/runbook headers) and shipped reality. Use when docs have lagged weeks of shipping, after a large multi-wave effort completes, or when the user says 'update/audit all the docs'. Works across one or many repos.

DevOps & Infrastructure Listed

pr-ci-cd-ship

Drive a code change hands-off through PR → CI → merge → (optionally gated) deploy → live verify, self-pacing across long-running CI/deploy with background run-watches instead of polling. Use whenever shipping a branch through a CI/CD pipeline that has watchable runs and possibly a manual approval gate. Stack-agnostic (gh/GitLab/etc.).

verify-live-integration

Verify code against the REAL external dependency, not just mocks — unit mocks encode the API shape you ASSUMED, so they pass green while production fails. Use before/after shipping any integration with a third-party or beta SDK/service, and when building destructive ops (deletes, archives, migrations, bulk mutations) that must be safe to run live.

fleet

Run an agent-fleet wave on the current repo — Claude (Fable) acts as boss, decomposes a goal into file-disjoint tasks, fans out tiered subagents (opus/sonnet/haiku) in isolated git worktrees via the Workflow tool, adversarially verifies each result, then merges survivors serially so main stays green. Use when the user runs /fleet <goal> or asks to "run the fleet" / "fleet this".

Web & Frontend Listed

prototype-to-parity

Turn an APPROVED prototype (clickable mockup, design comp, Figma, an HTML/JS prototype) into the shipped product 1:1 — extract a faithful parity spec, implement test-first, then adversarially review the build AGAINST the prototype before shipping. Use whenever the user has signed off on a prototype/mockup and wants the real implementation, or says the implementation 'doesn't look like the prototype'. Hand the ship step (PR→CI→deploy→verify) to your CI/CD + deploy skills.

sn-aia-agents

Use BEFORE building, wiring, or debugging ServiceNow AI Agent Studio agents (sn_aia_*) — script-tool contract, trigger/use-case wiring, run-as identity, channels, and the failure-signature playbook. Hard-won on a live dev instance (2026-08); every rule here cost hours to learn.

sn-atf

Use BEFORE running or authoring ServiceNow ATF tests against a live dev instance — browser test runner requirement, suite creation, clean-state discipline, trigger interplay, and result polling. Hard-won on a live dev instance, 2026-08.

Code & Development Listed

sn-live-debug

Use when debugging anything on a live ServiceNow instance — server-side script probing via an MCP script runner, schema discovery, syslog technique, evidence-first methodology (diff a working analog, instance-wide aggregates), and safe test-data fabrication. Distilled from the 2026-08 harness debugging session.

API & Backend Listed

sn-scripting

Use BEFORE writing any ServiceNow server/client script (GlideRecord, Script Includes, Business Rules, .server.js/.client.js, Jelly UI Pages) — required by the authoring gate. Documents the scoped-API allow-list, silent-failure patterns, and Jelly escape rules.

Data & Documents Listed

sn-sdk-fluent

Use BEFORE writing or editing any ServiceNow Fluent DSL file (.now.ts) — required by the authoring gate. Documents Fluent statement restrictions, primitive types, $id/Now.ID rules, reference integrity, and the live-docs oracle.