Interrogate

When you need a second opinion on a code change but don't trust a single reviewer's blind spots, this spawns multiple LLMs (Opus, GPT, Grok by default) to tear apart your diff from independent angles. Each model gets the same prompt and rubric, so the adversarial pressure comes from model diversity, not roleplay. Consensus findings get flagged as high confidence, lone wolf critiques get noted but downweighted. The synthesis categorizes everything into act on, consider, noted, or dismissed, with rationale for each bucket. It's explicitly opinionated: you're the lead reviewer with full context, and the skill backs aggressive filtering of noise. Good for changes where the cost of missing something is high.

skills.sh installs: 5.3K
GitHub stars: 15.9K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add cursor/plugins --skill interrogate --agent claude-code

Installs into .claude/skills of the current project.

Files

Interrogate

Adversaries try to break a change or a design. A finding survives only when a run proves it. Report the findings. Do not apply fixes.

Frame

The scope is what the user names. Otherwise it is the branch diff from its merge base with the base branch, through the working tree. State the intent with the in-your-own-words skill, from the user's message, the commits, and the PR description. Ask the user when the intent is unclear.

Fan out

Run Phase A steps 3 and 4 and Phase B of the arena skill, with four changes. Read the interrogate reviewers line in place of arena runners. Copy the scope's uncommitted changes into each worktree. Spawn each adversary as generalPurpose. In place of the arena task and rationale, give each adversary the scope, the intent, and this brief:

SKILL.md on GitHub