Agent
Fresh-context adversarial reviewer for a freshly-authored .claude/ artifact (skill, hook, or agent). Dispatched by /author's design-review gate after fill, before verify. Attacks the artifact's DESIGN — wrong step graph, missing gate, prose describing a deterministic step, a description that will never fire, unhandled…
Agent
Fresh-context adversarial reviewer — catches blind spots hot-context self-review cannot. Dispatched by /implement-audit skill for the core review work. Input is a curated brief (commit SHA range, changed-files list, project context, optional SPEC). Output is triaged findings (must-fix / should-fix / nit) in a…
Agent
Fresh-context in-repo investigator. Dispatched by /explain (and any skill needing a focused deep-dive that shouldn't pollute the main context). Returns structured findings with exact file:line citations so the caller can synthesize without carrying the search traffic. Single-round, cannot self-invoke.
Agent
Fresh-context reviewer that checks whether compressing a rule dropped any behavior. Dispatched by /author's efficacy-gate step during rule compression. Receives the original rule text, the compressed rewrite, and the list of dropped clauses — blind to the author's rationale — and returns a per-clause verdict of…
Agent
Fresh-context reviewer that checks whether a SPEC's interpretation matches the user's verbatim ask. Reads ONLY the user's literal words + the SPEC's Behavior section — nothing else. Flags interpretation drift (e.g., user said "function", SPEC describes "deployed service"). Dispatch it after writing a SPEC, before it…
Skill Claude CodeCodex
Scaffold a conforming AI-layer artifact — a rule, skill, hook, agent, or output style — that already passes the format checks, then verify it. Use when adding any new .claude/ artifact, so it matches the existing shape instead of drifting.
Skill Claude CodeCodex
The "wrap" verb of the loop — end the session by writing the journal entry, updating touched project READMEs, committing, and pushing. Auto-invoked as the tail of /spec and /implement (the producer verbs) on clean completion, so "wrap" is not a verb the user types; also runnable directly. Dispatches /implement-audit…
Skill Claude CodeCodex
Adversarial debate — spawn competing AI agents to argue a question, then synthesize the strongest conclusion. Use when the user says "debate this", "I'm torn between X and Y", "red-team this plan", or "council on this".
Skill Claude CodeCodex
Deep research into a topic or issue — investigate until confident, then present executive summary and root cause analysis. Use when you need to understand WHY.
Skill Claude CodeCodex
Adversarial review of freshly-built code — find what was missed, what's inconsistent, what breaks. The single code reviewer for the loop, and it runs exactly once per session. Auto-fired by /implement; standalone-callable after any substantial ad-hoc change. Emits the implement-audit: commit trailer the git hook…
Skill Claude CodeCodex
The "do" verb of the loop — execute a lean SPEC with rigorous self-validation. Use ONLY in a fresh context, after a prior session wrote the SPEC via /project.
Skill Claude CodeCodex
Single entry point for project work. No args renders the live priority-sorted project index. Existing slug loads the README, detects stage via scripts/detect-stage.sh, and routes to the right action (think flow, /implement, /close, or report-state). New freeform creates a project and runs the think flow.…
Skill Claude CodeCodex
Adversarial review of a freshly-written or materially-rewritten SPEC, before any code is written. Runs an independent reviewer against the SPEC's design and the spirit-check agent against the user's verbatim ask, in one pass. Auto-fired by /spec; also callable standalone after any SPEC edit. Emits the spec-audit…
Skill Claude CodeCodex
Write the SPEC. Given a design (handed over by /project, or gathered ad-hoc mid-session), write the lean 4-section SPEC file, audit it via /spec-audit (independent design review + spirit-check), then auto-invoke /close on a clean pass. The SPEC-writing half of the loop's Think verb — broken out so a SPEC can be…
Command
Wrap — journal what happened and why, then commit.
Command
Review — adversarially check a freshly-built change before it lands.
Command
Do — execute the SPEC in a fresh context, validating each step.
Command
Think — plan the work and write a lean SPEC before building.
Command
Review — attack the SPEC before any code is written.
Command
Think — write the lean SPEC that the build will be measured against.