Skill Claude CodeCodex
Use BEFORE designing any new way to observe, report, or coordinate something in the ai-conductor repository — a watcher, a poller, a sidecar file, an ad-hoc log, a second telemetry path, or a timestamp stamped into an artifact to be read back later. Also use when adding a member to the ConductorEvent union…
Skill Claude CodeCodex
Review and maintain this repository's human-facing documentation. Use when this repository invokes its maintain-documentation custom step or explicitly requests documentation maintenance.
Skill Claude CodeCodex
Judge this repository's implementation diff and write its authoritative structured release disposition to the retained SHIP draft PR before finish.
Skill Claude CodeCodex
Use before authoring any change to the ai-conductor harness repository, and before creating any new skill, to decide three things deterministically: whether the change is harness-repo-only or consumer-facing, whether a new skill belongs in the shipped skills/ catalog or this repository's local .agents/skills/ catalog…
Skill Claude CodeCodex
Use whenever adding, changing, reviewing, or debugging tests in the ai-conductor repository. Defines repository-specific test scope, isolation, mocking, fixture, performance, and CI-parity rules that prevent cyclic Conductor runs, leaked workers, real third-party calls, and slow aggregate suites. Complements the…
Skill Claude CodeCodex
Use only within an active harness lifecycle when its architecture-diagram step is current. Generates and maintains C4 Mermaid diagrams; do not invoke for ordinary code changes or general architecture questions.
Skill Claude CodeCodex
Use only within an active harness lifecycle when pre-stories feasibility review, a pipeline batch drift check, or the SHIP as-built review is current. Do not invoke for ordinary implementation, code review, or general architecture questions.
Skill Claude CodeCodex
Use for codebase health assessment. Dispatches 9 specialist agents + CTO orchestrator to evaluate security, data integrity, dependencies, architecture, duplication, testing, infrastructure, observability, and developer experience.
Skill Claude CodeCodex
Use when starting a new project, onboarding to an existing project, or setting up the harness for the first time. Detects project type, tech stack, and generates project-specific configuration.
Skill Claude CodeCodex
Judge whether criterion-bound changed tests can pass without the behavior they claim to cover.
Skill Claude CodeCodex
Use after implementing a task, before merging, or when requesting quality verification. Dispatches an evaluator agent with fresh context for calibrated, skeptical review.
Skill Claude CodeCodex
Use only within active engineer/conduct DECIDE after plan for a Medium or Large spec. Authors the committed outcomes-to-requirements-to-stories-to-tasks mapping; do not invoke for Small work or ordinary traceability questions.
Skill Claude CodeCodex
Interactive, phone-drivable idea→spec loop: hands a raw idea to the right repo, runs the full DECIDE phase there, and opens a spec PR. Use when capturing and routing new work, NOT when building inside one repo (that is plain conduct).
Skill Claude CodeCodex
Use to guide a feature through the full SDLC. Checks artifact state, determines current phase, tells you what to run next, and blocks progression when gates aren't met.
Skill Claude CodeCodex
Use only within active engineer/conduct DECIDE after accepted stories and before plan, or when that workflow explicitly rechecks remediated stories. Detects story interactions; do not invoke for Git conflicts, merge conflicts, or general requirements discussion.
Skill Claude CodeCodex
Use when a feature is stuck in daemon execution — halted, spinning, stalled, or silently not progressing — and an operator needs to know why. Diagnoses the failure and routes it to the right runbook. Operator-invoked only; never auto-dispatched, and never mutates anything without explicit per-action approval.
Skill Claude CodeCodex
Use only when the requested task is diagnosis of an observed bug, test failure, or unexpected behavior, or when active BUILD encounters one. Do not invoke for feature implementation, proactive review, or speculative troubleshooting without a concrete failure.
Skill Claude CodeCodex
Deprecated compatibility delegate for the canonical composer idea→spec workflow. Use when an existing /engineer or $engineer invocation must continue to work.
Skill Claude CodeCodex
Use only within an active engineer/conduct DECIDE workflow when approach selection and product-versus-technical track classification are still unresolved. Do not invoke for ordinary implementation, review, explanation, or an already-decided change or hotfix.
Skill Claude CodeCodex
Use at the FINISH boundary to gather an operator's publication intent, author the retained PR's reader-facing prose when it is still unauthored, and judge that prose; the engine-owned publication coordinator performs all deterministic publication mechanics.
Skill Claude CodeCodex
Use only when the requested deliverable is a GitHub intake issue, or when an active harness workflow hands an observation off for later DECIDE. Do not invoke merely because a bug, idea, or observation is being discussed.
Skill Claude CodeCodex
Use after /finish to validate stories via curl (API) or browser (full-stack). Browser capability gaps warn; observed bugs loop back through /tdd.
Skill Claude CodeCodex
Use when the operator requests project recall, when the harness dispatches its memory step, or when an active harness workflow requires a memory checkpoint. Recall-before-act protocol with categorized persistence and staleness detection.
Skill Claude CodeCodex
Use when executing an implementation plan with multiple tasks. Factory orchestration with three autonomy levels, quality gates, rework budgets, and audit trails.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: