Subagent Prompts

GPTClaudeDeepSeek··▲ 230 copies·updated 2026-07-14

subagent-prompts-4.prompt

# Subagent Prompts

Use these after the main skill framework exists to get independent critique.

## Autoimmune Reviewer

Review the Agent Immune System skill as an autoimmune detector. Find places where it could create more process, token burn, or overcorrection than it prevents. Return only concrete risks and recommended edits.

## Promotion Reviewer

Review the Promotion Rubric. Identify any path where a one-off incident could still become active global instruction memory. Recommend tighter gates.

## Regeneration Reviewer

Review the Regeneration Rubric and templates. Find whether the workflow actually compresses scar tissue into principles, or merely creates another report. Recommend changes that make simplification measurable.

## Hook Reviewer

Review the Claude and Codex hook scripts. Confirm they do not mutate active instruction files, write only minimal audit logs, stay low-noise, and are safe to run often. Identify any risk of blocking normal work or logging sensitive content.

when to use it

Community prompt sourced from the open-source GitHub repo inbusiness23/agent-immune-system (MIT). A "Subagent Prompts" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.