Agent
The AUDITOR (codex voice). THE PROCEDURE step 6 (VERIFY), routed to Codex per Ro's directive 2026-08-02 ("queremos usar Codex más porque nos dan más créditos"). Reads a diff or finished unit against its spec, reports findings, and stops. It NEVER edits and NEVER spawns anything — this agent has no Write/Edit tools at…
Agent
The BUILDER. THE PROCEDURE step 5 (BUILD) — dispatches ONE unit to the real synchronous Codex CLI (codex exec), which edits files on disk, and returns the diff. This agent has NO file-editing tools itself; it cannot write code in Claude even by accident. This is NOT codex:codex-rescue — that plugin agent is a…
Agent
Delegates work to REAL Gemini 2.5 (Pro/Flash) via the gemini shim → OpenCode. A genuine independent model, so valid as the non-Claude voice in a council. Gemini gets its OWN Read/Bash/Edit tools inside OpenCode, so it can scrape, run commands, read repos, and analyse images — not just answer from context. Use for…
Agent
Runs after a unit of work is built and verified, to make every code map tell the truth about the repo again. THE PROCEDURE step 7 (PERSIST), before the commit. It regenerates the CACHED maps (.codemap, graphify) and VERIFIES the LIVE ones (L0/L1/skeleton) — those are computed on every call and have nothing to…
Agent
The AUDITOR. THE PROCEDURE step 6 (VERIFY) — a non-builder reviews the builder's work before it is accepted. Reads a diff (or a finished unit) against its spec, reports findings, and stops. It NEVER edits and NEVER spawns anything. Use for: auditing a codex diff; checking a unit against its CONTEXT/CHANGE/GOAL/VERIFY…
Skill Claude CodeCodex
Activate Ro's context-light coding procedure, evidence-led tool routing, bounded builder/auditor delegation, verification, and persistence. Use when Ro invokes awesomeharness or /awesomeharness, or for substantive implementation, refactors, and audits in an awesome-harness repository.
Skill Claude CodeCodex
Run a deterministic battery of pre-ship checks against any repo — composes the repo's own gate commands and adds universal ones (file-size caps, no-TODO scan, duplicate-code detection, optional semgrep). Use before shipping, before handing work back to Ro, before opening a PR, or whenever Ro says "check-all" / "run…
Skill Claude CodeCodex
The standard coding workflow for a Codex session — decompose a code change into small, independently-verifiable units BEFORE writing any of it, then execute and self-audit them one at a time. Use whenever you are about to write or change code beyond a one-line edit: features, refactors, bug fixes, new modules…
Skill Claude CodeCodex
Prove what an existing repository already provides before adding a feature, module, integration, dependency, helper, or material refactor. Return evidence-backed REUSE/ADAPT/REJECT decisions and a STOP/PLAN/BUILD gate before code-decompose.
Skill Claude CodeCodex
The every-turn close for a Codex session — commit the work, record the durable lesson, update the STATE resume point and .now.md, push, and emit the CONTINUE block. Run at the end of every substantive turn and before any context compaction, so nothing that only lives in the conversation is lost. Works in any repo.
Skill Claude CodeCodex
Retrieve durable memory fast instead of grepping scattered markdown. Use when you need prior context on a topic ("what do we know about X", "have we decided Y"), at the start of substantive work to load task-relevant memory, or before proposing something that may already be recorded. Backed by the memgraph full-text…
Skill Claude CodeCodex
Turn a fuzzy objective into a running, self-verifying loop. Use when Ro wants to "create a loop", run something "until it's done", or hand off a multi-step objective instead of hand-prompting each step ("/goal", "loop until", "keep going until tests pass", "build X autonomously"). The skill manufactures a strong SEED…
Skill Claude CodeCodex
Recurring harness intelligence pass, proposal-only, two modes. (A) AUDIT — map one repo's harness (CLAUDE.md, front-door docs, mulch, STATE, agent specs) against the REAL codebase and produce a fidelity/behavioral drift report. (B) SCOUT — mine Ro's recent transcripts for repeated toil to automate, and research…
Skill Claude CodeCodex
Procesar el Video Inbox de Notion — transcribir links de videos pegados por Ro, clasificarlos por proyecto, y rutearlos a Notion + espejo local /Notes. Activar cuando Ro diga "procesa el inbox", "checa el inbox", "/notes-inbox", o pegue un link de video pidiendo transcripción/ruteo.
Skill Claude CodeCodex
THE PROCEDURE steps 1-3 in ONE call — RECALL (memgraph + mulch + MEMORY.md) then UNDERSTAND (front door, graphify + repowise, ponytail reuse ladder) then the REUSE/ADAPT/REJECT verdict and STOP/PLAN/BUILD gate. Run before writing code, before proposing something that may already exist, or when a request would add a…
Skill Claude CodeCodex
Debug web UIs live in Ro's Chrome — read console errors, reload, screenshot, iterate until zero errors. Use for any "no se ve nada / está roto / está trabado" report about a local web page.
Skill Claude CodeCodex
Research credible creators and mine their content on YouTube + Instagram/Reels for any niche. Use when the user says "run the youtube research skill", "do youtube research", "research on youtube", wants to find high-credibility creators/channels on a topic, pull a video's transcript, or scrape Instagram reels/comments…