Claude Second Opinion

VerifiedCaution

Get an independent review from Claude Code on your code or plans. Useful for cross-checking reasoning from Codex.

Sby Skills Guide Bot
DevelopmentIntermediate
7/24/2026
CodexClaude Code
#second-opinion#code-review#cross-check#claude-code#read-only

Recommended for

Automated summary

Delegates a question or review task to Claude Code in non-interactive print mode and returns its independent assessment.

Strengths

  • Builds self-contained, unbiased prompts with explicit file context and read-only constraints.
  • Uses a restricted read-only tool set (Read, Grep, Glob, LS) to avoid unintended modifications.
  • Includes troubleshooting for authentication, tool restrictions, and truncated output.
  • Reports Claude's answer alongside agreement or disagreement for the user to decide.

Limitations

  • Requires a locally installed and authenticated Claude CLI; otherwise it only reports the blocker.
  • The read-only tool restriction may prevent Claude from running tests or observing runtime behavior.
  • Long or open-ended reviews can be truncated, so prompts should stay focused.
When to use it

Use when you want an independent cross-check of code, a plan, or reasoning without letting the second agent modify files.

When not to use it

Do not use when you need Claude to implement changes, run tests, or when the Claude CLI is unavailable or unauthenticated.

Security analysis

Caution
Quality score85/100

No destructive or exfiltrating behavior: the skill shells out only to the `claude` CLI with a read-only tool allowlist, explicitly forbids file modification, and tells the agent to preserve read-only restrictions. The main risks are ordinary ones for AI-delegation skills — network/credential use, possible shell-quoting issues when embedding user task text, and reliance on a non-interactive permission mode. These warrant 'caution' rather than 'safe'.

Findings
  • •Invokes an external AI CLI (`claude -p`) as a subprocess, which spends tokens/credentials and makes network calls to Anthropic
  • •Prompt/argument interpolation into a shell command (`"<prompt>"`, `$ARGUMENTS`) could break quoting if the task text contains quotes or shell metacharacters; the skill should require proper escaping rather than raw interpolation
  • •Uses `--permission-mode dontAsk`, which suppresses interactive permission prompts; mitigated here because `--allowedTools` is restricted to Read,Grep,Glob,LS, but the pattern is worth flagging since a weaker model might drop that restriction
How this analysis is made

Examples

Review retry logic
Ask Claude for a second opinion on whether the retry logic in src/config.py is correct. Have it read the file and do not modify anything.
Validate a plan
Get Claude's independent review of this migration plan: [paste plan]. Ask whether any steps are risky or missing.
Cross-check architecture
Use Claude to cross-check my proposed caching architecture in docs/cache-plan.md and list concerns without changing files.

name: claude description: Get a second opinion from Claude Code. Use from Codex when you want to cross-check your reasoning, validate a plan, or get an independent review from Claude. argument-hint: <task or question for Claude> user-invocable: true allowed-tools: Bash(claude *), Read, Grep, Glob

Claude Second Opinion

Delegate a task to Claude Code for an independent review or second opinion.

Input

<task> $ARGUMENTS </task>

If no task is provided, ask the user what they want Claude to review.

How to Use

Run Claude in non-interactive print mode from the current working directory. Prefer a read-only tool set:

claude -p \
  --permission-mode dontAsk \
  --allowedTools Read,Grep,Glob,LS \
  --add-dir "<current working directory>" \
  "<prompt>"

If the Claude CLI is not installed or is not authenticated, report that blocker and include the command output.

Prompt Construction

Build a self-contained prompt for Claude that includes:

  1. Context: The repo path, what project this is, and what folder/files are relevant
  2. The task: What you want Claude to do (review, validate, suggest, etc.)
  3. Constraints: Tell it to only read files, not modify anything

Keep it unbiased. Do not include your own conclusions or preferred answer. Present the problem neutrally so Claude forms its own opinion.

Be specific about files. Tell Claude exactly which files to read — it has access to the repo but needs direction.

Example

claude -p \
  --permission-mode dontAsk \
  --allowedTools Read,Grep,Glob,LS \
  --add-dir "/path/to/project" \
  "Review the file at src/config.py. Is the retry logic correct? Read the file and give your assessment. Do NOT modify any files."

Troubleshooting

  • Authentication: If Claude exits with an auth error, ask the user to run claude auth or configure the required Claude credentials before retrying.
  • Tool restrictions: If Claude cannot read the requested files, retry with a narrower prompt and the same read-only tools. Do not remove read-only restrictions unless the user explicitly approves.
  • Output truncation: Claude may produce a long review. Keep prompts focused and ask specific questions rather than open-ended reviews.
  • Session state: Use -p for a fresh non-interactive answer. Do not use --continue or --resume unless the user specifically asks to continue an existing Claude session.

After Claude Responds

  1. Report the Claude response to the user
  2. Note where you agree or disagree with Claude's assessment
  3. Let the user decide how to proceed

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/claude-second-opinion-mryiqgun/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/claude-second-opinion-mryiqgun' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/claude-second-opinion-mryiqgun/download' --output '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md.part' && mv '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md.part' '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

Tools declared by the author (check availability in your agent): Bash(claude *), Read, Grep, Glob

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md' && test -s '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/claude' && cp -R '.skill-downloads/claude-second-opinion-mryiqgun' '.claude/skills/claude'

Codex
test -f '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md' && test -s '.skill-downloads/claude-second-opinion-mryiqgun/SKILL.md' && mkdir -p '.agents/skills' && test ! -e '.agents/skills/claude' && cp -R '.skill-downloads/claude-second-opinion-mryiqgun' '.agents/skills/claude'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/claude-second-opinion-mryiqgun'

openskills read 'claude-second-opinion-mryiqgun'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/claude-second-opinion-mryiqgun/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills