# Waki285 Dotfiles Codex Consult

> Codex Consult

- Skill: `tomevault-io/waki285-dotfiles-codex-consult` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add tomevault-io/waki285-dotfiles-codex-consult`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/waki285-dotfiles-codex-consult/raw
- Safety review: pending (external: skill-scanner PASS, skillspector CAUTION)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: tomevault-io (https://skillmd.com/u/tomevault-io)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/tomevault-io/waki285-dotfiles-codex-consult

---


# Codex Consult

## Auto-trigger Guidelines

Proactively consult Codex when:
- Stuck on an obscure error after 2+ failed attempts
- Facing an architectural decision with significant trade-offs
- Working with an unfamiliar library/API not covered by Context7
- The user's problem has multiple valid solutions and a second perspective would help
- **After completing a plan in plan mode** — consider consulting Codex to validate the approach, catch blind spots, or get alternative perspectives before presenting the plan to the user

Do NOT auto-trigger when:
- The answer is straightforward or well-known
- Context7 or web search already provides sufficient information
- Already consulted Codex in the current conversation for the same topic

## Execution

Run `codex exec` directly via the Bash tool. Do NOT use wrapper scripts.

**Timeout**: Codex can take up to 10 minutes or more for complex queries. Always set the
Bash tool's `timeout` parameter to `600000` (10 minutes) when invoking `codex exec`.

The output directory `~/.claude/second-opinions/` should already exist. If it does not,
the `-o` flag will create it implicitly via `codex exec`.

### Step 1: Run with ym profile (primary)

```bash
codex exec -p ym --full-auto --skip-git-repo-check -o ~/.claude/second-opinions/<TIMESTAMP>_<TOPIC>.md "<PROMPT>"
```

- `<TIMESTAMP>`: Use `$(date +%Y-%m-%d_%H-%M-%S)` format
- `<TOPIC>`: Lowercase, alphanumeric + hyphens only, max 30 chars (e.g., `rust-lifetime`, `api-design`)
- `<PROMPT>`: The consultation prompt (see Prompt Formulation below)

### Step 2: Fallback (if Step 1 fails)

If the voids provider errors, retry without `-p ym`:

```bash
codex exec --full-auto --skip-git-repo-check -o ~/.claude/second-opinions/<TIMESTAMP>_<TOPIC>.md "<PROMPT>"
```

## Prompt Formulation

Craft prompts with this structure:

```
Context: [language, framework, versions]
Problem: [clear description]
Code: [relevant snippet, under 100 lines]
Question: [specific, open-ended question]
```

Keep prompts focused. Summarize large code blocks — do not dump entire files.

## Response Handling

1. Read the saved output file with the Read tool
2. Synthesize Codex's response with your own analysis
3. Present to the user:
   - Mention that Codex was consulted
   - Show the saved file path
   - Highlight where perspectives align or differ
   - Give a combined recommendation

## Security

Never include in prompts:
- API keys, tokens, passwords, or secrets
- Personal identifiable information
- Proprietary business logic (unless the user explicitly permits it)

---
> Converted and distributed by [TomeVault](https://tomevault.io/claim/waki285) — claim your Tome and manage your conversions.
<!-- tomevault:4.0:skill_md:2026-04-11 -->