Menú
Use when the user wants to delegate a task to OpenAI Codex CLI, get a second opinion, or run codex review on changes
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
Save current session state to a handover document for seamless continuation in a new session. Invoke manually or triggered automatically when context window reaches 85%.
Configurable 9-phase feature development workflow enforcer (Phase 0 Context Setup → Phase 8 Approval & Commit). Delegates planning to OpenSpec (proposal→specs→design→tasks) and enhances execution with superpowers skills (TDD, subagent-driven-development, worktrees, verification). Auto-validates phase transitions, manages state via context.md, and orchestrates specialized agents. Use when user says "start feature" or "resume feature", mentions implementing/building/creating a feature, needs structured development process, references TDD or test-driven development, asks about feature planning or implementation phases, references 9-phase workflow or mentions "Phase 1" or "Planning phase", needs approval gates or phase validation, or asks "how do I start a feature" or "what's the development workflow".
Structured E2E test troubleshooting using the 6-layer diagnostic ladder (Frontend → API → SP/Schema → Infra → CI/CD). Use when any E2E test fails, UI doesn't render, auth drops, unexpected status codes, CI passes locally but fails in pipeline, or any test failure where the root cause is unclear. Routes to the correct read-only diagnostic agent based on the symptom, walks the sequential ladder with [EVIDENCE]/[HANDOFF]/[ROOT-CAUSE] taxonomy, and enforces the Fix Gate (no action before root cause confirmed).
Post-session retrospective — append tagged learnings (with severity) to configured learnings file (default: `learnings.md`)
SQL Server database development specialist. Use when creating tables, stored procedures, migrations, or troubleshooting SQL Server/Docker issues. Reads project tech-stack directory conventions if present; falls back to generic SQL Server patterns (UNIQUEIDENTIFIER PKs, CreatedOn/ModifiedOn with GETUTCDATE(), no FK constraints, WITH(NOLOCK), SARG optimization, stored proc naming {Product}_{Module}_{Function}_{Date}).
| name | codex |
| description | Use when the user wants to delegate a task to OpenAI Codex CLI, get a second opinion, or run codex review on changes |
Delegate tasks to OpenAI Codex CLI from within Claude Code.
| User says | Resolves to |
|---|---|
/codex review | codex review --uncommitted |
/codex review --base main | codex review --base main |
/codex review with o3 effort: high | codex review --uncommitted -m o3 -c model_reasoning_effort="high" |
/codex security | codex review --uncommitted -c model_reasoning_effort="high" "Focus on security: injection, auth bypass, data exposure, OWASP Top 10" |
/codex perf --base main | codex review --base main -c model_reasoning_effort="high" "Focus on performance: N+1 queries, unnecessary allocations, missing indexes, expensive loops" |
/codex quick explain this error | codex exec --full-auto -c model_reasoning_effort="low" "explain this error" |
/codex thorough refactor auth | codex exec --full-auto -c model_reasoning_effort="xhigh" "refactor auth" |
/codex fix | Capture git diff + git diff --cached → codex exec --full-auto "Review and fix issues in this diff: ..." |
/codex fix the broken test | codex exec --full-auto "fix the broken test" |
/codex second-opinion | Rerun last user prompt through Codex, present both answers |
/codex <anything else> | codex exec --full-auto "<prompt>" |
digraph routing {
"Parse input" [shape=ellipse];
"Preset?" [shape=diamond];
"second-opinion?" [shape=diamond];
"Bare fix?" [shape=diamond];
"Review keyword?" [shape=diamond];
"Expand preset" [shape=box];
"Second opinion flow" [shape=box];
"Diff-scoped fix" [shape=box];
"codex review" [shape=box];
"codex exec" [shape=box];
"Parse input" -> "Preset?";
"Preset?" -> "Expand preset" [label="yes"];
"Preset?" -> "second-opinion?" [label="no"];
"second-opinion?" -> "Second opinion flow" [label="yes"];
"second-opinion?" -> "Bare fix?" [label="no"];
"Bare fix?" -> "Diff-scoped fix" [label="yes"];
"Bare fix?" -> "Review keyword?" [label="no"];
"Review keyword?" -> "codex review" [label="yes"];
"Review keyword?" -> "codex exec" [label="no"];
}
Review keywords: review, code review, check changes, audit, inspect, critique
Default: Everything else routes to codex exec --full-auto
After routing, for all commands: extract model/effort flags → inject project context (review + presets only) → execute via Bash → summarize results.
Detect as first token after /codex. Additional text appends to the prompt. User can override model/effort.
| Preset | Mode | Effort | Focus prompt |
|---|---|---|---|
security | review | high | Security: injection, auth bypass, data exposure, OWASP Top 10 |
perf | review | high | Performance: N+1 queries, unnecessary allocations, missing indexes, expensive loops |
quick | exec | low | (user's remaining prompt) |
thorough | exec | xhigh | (user's remaining prompt) |
Trigger: /codex second-opinion or /codex second opinion
codex exec --full-auto "<original prompt>"Append custom focus: /codex second-opinion focus on error handling
Trigger: /codex fix with NO additional prompt text
git diff + git diff --cached — if empty, stop with "No uncommitted changes"codex exec --full-auto "Review the following diff and fix any bugs, logic errors, or code quality issues:\n\n<diff>"/codex fix <something> (with a prompt) routes to normal exec mode instead.
Extract model/effort from user input before building the command. Strip from the prompt.
| Input pattern | Flag produced |
|---|---|
-m o3 / --model o3 / with o3 / using o3 | -m o3 |
--effort high / --thinking high / effort: high | -c model_reasoning_effort="high" |
Valid levels: low, medium, high, xhigh
Default: Inherit from ~/.codex/config.toml
Applies to: review commands and preset commands only (not plain exec).
Read CLAUDE.md from project root. Extract key conventions (~200 words max) and prepend:
Project conventions: <framework>, <DB>, <patterns>.
<extracted rules>
---
<original prompt>
Focus on rules Codex would otherwise violate: architecture layers, DB conventions, code style.
--uncommitted and --base are mutually exclusive — default to --uncommitted-m, --base, -c, etc.) verbatim