Agent skill

Co Validate

by SnakeO in SnakeO/claude-co-commands

Get a staff engineer review of your plan via Codex. An agent skill from SnakeO/claude-co-commands.

MITAuto-check passedAgent Workflows

Install Co Validate

skills CLI
$ npx skills add SnakeO/claude-co-commands --skill co-validate -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install SnakeO/claude-co-commands co-validate --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/SnakeO/claude-co-commands.git skills-src && mkdir -p .claude/skills && cp -r skills-src/plugins/co-commands/skills/co-validate .claude/skills/co-validate && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
co-validate
GitHub stars
148
Token cost
~1.5k tokens
SKILL.md length
683 words
Files
1
Skills in repo
3
Repo updated
First seen
Licence
MIT

At a glance

Get a staff engineer review of your plan via Codex. An agent skill from SnakeO/claude-co-commands.

  • Works in 3 steps: Start Codex and Hand It to a Background… → Do Your Own Review → Retrieve and Compare
  • You want critical review feedback on a plan before finalizing it
  • SKILL.md covers Arguments, How Codex Is Reached, Step 1: Start Codex and Hand… and Step 2: Do Your Own Review, plus 3 more sections
  • Calls go, claude and codex

What it does

Co Validate is an agent skill from SnakeO/claude-co-commands. Get a staff engineer review of your plan via Codex. Use when you want critical review feedback on a plan before finalizing it. Pass the path to the plan file as the argument.

Its SKILL.md is about 1.5k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows. It works with Bash. The repository describes itself as: Co- planning, validation, and brainstorming commands for Claude Code via Codex MCP. The licence is MIT.

When your agent uses it

  • You want critical review feedback on a plan before finalizing it

Example prompts

  • “/co-validate”

Workflow steps

3 steps, taken from the step headings in SKILL.md.

  1. Start Codex and Hand It to a Background Subagent
  2. Do Your Own Review
  3. Retrieve and Compare

What it can do on your machine

Read from SKILL.md and the folder at commit 7c477b2. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • go
    • claude
    • codex

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Co Validate loads about 1.5k tokens when it runs. Until then it costs about 47 tokens; SKILL.md has 683 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~47
When it runs · the whole SKILL.md, loaded when a task matches
~1.5k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from SnakeO/claude-co-commands at commit 7c477b2, republished under its MIT licence (© SnakeO). 683 words, ~1,473 tokens.

Download SKILL.mdSave it as .claude/skills/co-validate/SKILL.md (or your agent's skills folder).
name
co-validate
description
Get a staff engineer review of your plan via Codex. Use when you want critical review feedback on a plan before finalizing it. Pass the path to the plan file as the argument.

Co-Validate: Get a Staff Engineer Review of Your Plan

Arguments

$ARGUMENTS should be the path to the plan file. If not provided, check if there is a plan file from the current session (for example in ~/.claude/plans/ or the working directory).

How Codex Is Reached

Codex runs through its CLI, driven by the helper script scripts/codex-session, which lives two levels above this skill's base directory: call it as <base directory>/../../scripts/codex-session. It keeps one Codex conversation per session folder and always runs Codex in its read-only sandbox:

  • codex-session new <label> makes a session folder and prints its path
  • codex-session run <DIR> sends <DIR>/prompt.md as the first message and waits for the reply
  • codex-session read <DIR> prints Codex's latest reply
  • codex-session reply <DIR> sends a follow-up read from stdin and prints the reply

Shell variables do not survive between Bash calls, so always use the literal folder path.

Step 1: Start Codex and Hand It to a Background Subagent

Read the plan file, and find the original user prompt that led to the plan in the conversation history. Paste the original request between the tags below; the plan file itself is appended after the prompt with cat, not pasted.

  1. Create the session and write Codex's first message, in one Bash call:

    bash
    DIR="$(<base directory>/../../scripts/codex-session new co-validate)" && echo "$DIR"
    cat > "$DIR/prompt.md" <<'CODEX_PROMPT'
    You are a staff engineer reviewing this plan. Analyze it for critical issues, big simplifications, or a completely different better approach, reading the code in the working directory where it helps — but do NOT share your review yet. If you need to ask clarifying questions about the plan or original request before reviewing, ask them now. Otherwise, when your review is complete and fully formed, respond with exactly: "My review is complete and I'm ready to present" and nothing else. Wait for my next message before sharing your review.
    
    Original request from the user:
    <original_request>
    {paste the user's original prompt/request that triggered the plan}
    </original_request>
    
    Plan:
    CODEX_PROMPT
    cat "<path to the plan file>" >> "$DIR/prompt.md"
  2. Spawn a background subagent (Agent tool with run_in_background: true) to handle all communication with Codex. Give it the helper path and the folder path, and tell it to:

    1. Run <helper> run <DIR> in the foreground, with the Bash tool's longest timeout (600000 ms). It can take several minutes.
    2. Run <helper> read <DIR>. If Codex replied "My review is complete and I'm ready to present", go to step 4.
    3. If Codex asked clarifying questions instead, answer them using its own judgment and the code in the working directory, by sending a follow-up (quote the heredoc delimiter): <helper> reply <DIR> <<'CODEX_PROMPT' … CODEX_PROMPT. Then check the reply again as in step 2. If Codex sent its full review instead of the ready message, treat it as ready.
    4. Report back only that Codex is ready, the session folder, and how many questions it answered. Do NOT request or repeat Codex's review; that happens in Step 3.

    The subagent handles the back-and-forth so you are free to do your own work. In a non-interactive session (claude -p), do these steps yourself in the foreground instead: background work is killed when that kind of session ends.

Show full SKILL.md (278 more words)Show less

Step 2: Do Your Own Review

While the subagent communicates with Codex in the background, do your own independent review of the plan. Look for:

  • Critical issues or flaws in the approach
  • Opportunities for simplification
  • Missing edge cases or risks
  • Whether a completely different approach would be better

Write down your own assessment. Do NOT check the Codex result until you have finished your own review. The entire point is to produce two independent reviews and then compare them — reading Codex's review early defeats this purpose and introduces bias.

Step 3: Retrieve and Compare

Only after your review is complete, confirm the background subagent has reported that Codex is ready. Then ask Codex for its review in the same conversation; the command prints it:

bash
<base directory>/../../scripts/codex-session reply <DIR> <<'CODEX_PROMPT'
Go ahead, share your review. Be direct and concise. Do not repeat the plan back. Focus only on critical issues, big simplifications, or a completely different better approach.
CODEX_PROMPT

If the subagent reported that Codex did not reply, show the user the error it relayed (it names the fix, such as codex login) and continue with your own review alone.

Once the review arrives:

  1. Read the Codex review output.
  2. Compare it against your own review.
  3. For each issue raised (by either review), either:
    • Accept it and update the plan accordingly
    • Override it with an explanation of why the current approach is better

Continuing the Conversation

To address its points, explain overrides and ask for clarification, send a follow-up in the same Codex conversation. It prints Codex's answer:

bash
<base directory>/../../scripts/codex-session reply <DIR> <<'CODEX_PROMPT'
your response: points addressed, overrides explained, questions
CODEX_PROMPT

If you override points, explain why so Codex can push back if needed.

How to Treat Responses

Treat Codex responses as coming from a junior developer:

  • Never assume suggestions are correct; validate each one yourself.
  • You are the lead engineer and have final say.
  • Use responses as a starting point, not authoritative answers.

© SnakeO, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in plugins/co-commands/skills/co-validate of SnakeO/claude-co-commands.

Open the folder on GitHubat commit 7c477b2

Compare with similar skills

Co Validate next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Co Validate compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Co Validate this skillSnakeO/claude-co-commands148—~1.5kAutomated safety check: PassMIT
Hook Development for Claude Code Pluginsanthropics/claude-plugins-official38k11 repos~4.1kAutomated safety check: NotesApache-2.0
Plugin Settings Patternanthropics/claude-plugins-official38k7 repos~3kAutomated safety check: PassApache-2.0
Neat-Freak Knowledge CloseoutKKKKhazix/khazix-skills21k—~1.9kAutomated safety check: PassMIT
Crush Configurationcharmbracelet/crush29k—~3.7kAutomated safety check: PassCustom licence
Kimi Code DelegationCherryHQ/cherry-studio52k1 repos~504Automated safety check: PassAGPL-3.0

Similar skills

  • Hook Development for Claude Code Plugins

    anthropics/claude-plugins-official

    Official

    Explains how to write Claude Code plugin hooks, both prompt-based checks and bash commands, for events such as PreToolUse, Stop and SessionStart.

    38k GitHub starsUsed in 11 repos~4.1k tokens
    Agent WorkflowsAuto-check: notes
  • Plugin Settings Pattern

    anthropics/claude-plugins-official

    Official

    Shows how Claude Code plugins keep per-project settings and state in .claude/plugin-name.local.md files with YAML frontmatter and a markdown body.

    38k GitHub starsUsed in 7 repos~3k tokens
    Agent WorkflowsAuto-check passed
  • Neat-Freak Knowledge Closeout

    KKKKhazix/khazix-skills

    Brings project docs, agent rule files, authorized memory and leftover workspace files back in line with what the code and runtime actually do at the end of a work session.

    21k GitHub stars~1.9k tokensUpdated 6 days ago
    Agent WorkflowsAuto-check passed
  • Crush Configuration

    charmbracelet/crush

    Explains how to configure the Crush coding agent with crushrc or crush.json, covering providers, models, LSPs, MCP servers, hooks, permissions and config precedence.

    29k GitHub stars~3.7k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Kimi Code Delegation

    CherryHQ/cherry-studio

    Delegates one bounded repository task to Kimi Code in non-interactive prompt mode and reads back the final result from its JSON event stream.

    52k GitHub starsUsed in 1 repo~504 tokens
    Agent WorkflowsAuto-check passed
  • Official

    Designs and verifies a deterministic grader that measures whether a GitHub Agentic Workflow run reached its real-world or repository outcome.

    5.4k GitHub stars~6.8k tokensUpdated today
    Agent WorkflowsAuto-check passed

More from SnakeO/claude-co-commands

  • Co Brainstorm

    SnakeO/claude-co-commands

    Bounce ideas off Codex. An agent skill from SnakeO/claude-co-commands.

    148 GitHub stars~1.3k tokensUpdated 9 days ago
    Auto-check passed
  • Co Plan

    SnakeO/claude-co-commands

    Generate a parallel plan via Codex. An agent skill from SnakeO/claude-co-commands.

    148 GitHub stars~1.2k tokensUpdated 9 days ago
    Auto-check passed

Works with

Categories

Questions about Co Validate

What does Co Validate do?

Get a staff engineer review of your plan via Codex. An agent skill from SnakeO/claude-co-commands. Co Validate is an agent skill from SnakeO/claude-co-commands. Get a staff engineer review of your plan via Codex.

When should I use Co Validate?

Co Validate fits situations like: you want critical review feedback on a plan before finalizing it.

How do I install Co Validate in Claude Code?

Run `npx skills add SnakeO/claude-co-commands --skill co-validate -a claude-code`. Or copy the skill folder (plugins/co-commands/skills/co-validate in SnakeO/claude-co-commands) into .claude/skills/co-validate in your project. Claude Code loads it when a task matches its description.

How do I install Co Validate in Codex?

Run `npx skills add SnakeO/claude-co-commands --skill co-validate -a codex`. Or copy the skill folder (plugins/co-commands/skills/co-validate in SnakeO/claude-co-commands) into .agents/skills/co-validate in your project. Codex loads it when a task matches its description.

Can I use Co Validate in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add SnakeO/claude-co-commands --skill co-validate -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/co-validate, .gemini/skills/co-validate, .github/skills/co-validate and .opencode/skills/co-validate in your project.

What does Co Validate need to run?

Going by SKILL.md and its folder, Co Validate needs the command-line tools its instructions call (go, claude and codex).

Does Co Validate access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Co Validate safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Co Validate use?

Co Validate is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Co Validate use?

About 1.5k tokens (SKILL.md is roughly 5.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Co Validate?

Skills that share tags, products or a category with Co Validate: Hook Development for Claude Code Plugins (anthropics/claude-plugins-official, 38k stars), Plugin Settings Pattern (anthropics/claude-plugins-official, 38k stars), Neat-Freak Knowledge Closeout (KKKKhazix/khazix-skills, 21k stars) and Crush Configuration (charmbracelet/crush, 29k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Co Validate?

SnakeO (a GitHub user) maintains it in SnakeO/claude-co-commands, which has 148 GitHub stars. The repository holds 3 skills in this directory. The repository was last updated on September 28, 2026.

Source: SnakeO/claude-co-commands on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.