ask_skill

Overview

This skill lets you ask Claude, Codex, or Gemini from the local CLI and saves the response as a reusable artifact. It routes your prompt through the preferred local provider, captures the reply, and writes a timestamped Markdown record to the .omc artifacts folder. Use it to make outputs reproducible and shareable across your team.

How this skill works

You invoke the skill with the provider name and a question or task; it calls the corresponding local CLI (claude, codex, or gemini) and streams the provider output. The skill then persists the response to .omc/artifacts/ask/<provider>-<slug>-<timestamp>.md so the result can be referenced later. It requires the chosen CLI to be installed and authenticated on the machine.

When to use it

• You want to capture a canonical response from a specific local model provider.
• You need a shareable, timestamped record of a model review, plan, or suggestion.
• You are orchestrating multi-agent workflows and need agent outputs saved as artifacts.
• You want to compare outputs across Claude, Codex, and Gemini for the same prompt.
• You need an audit trail for decisions or code reviews generated by an AI.

Best practices

• Ensure the target CLI (claude, codex, or gemini) is installed and authenticated before invoking the skill.
• Run the provider --version command to verify availability in automation scripts.
• Keep prompts focused and include context like file paths, issue IDs, or diffs for reproducible results.
• Use the slug-friendly short description in the prompt to produce readable artifact filenames.
• Store artifacts in your repo or team storage for traceability and reviews.

Example use cases

• Run security-minded code review via: /oh-my-claudecode:ask codex "review this patch from a security perspective" and archive the result.
• Ask for implementation plans: /oh-my-claudecode:ask claude "draft an implementation plan for issue #123" and attach the artifact to the issue.
• Gather UX suggestions: /oh-my-claudecode:ask gemini "suggest UX improvements for this flow" and share the artifact with designers.
• Compare provider outputs for the same prompt to choose the best model for a task and keep the artifacts for later analysis.
• Automate team-runbook updates by querying a provider for a proposed change and saving the suggested edits as an artifact.