Interrogate (pstack)
Multi-model adversarial review: one read-only reviewer per configured model (default Claude + Grok) and a synthesized verdict; findings raised by several models rank higher. Does not auto-apply.
proof: logs blocks: agent grading its own homework
Source
https://github.com/cursor/plugins/tree/main/pstack/skills/interrogateChecked 2026-10-16. Read the original before installing.
By harness
| Harness | Status | How |
|---|---|---|
| Cursor | native | Native: ships in the pstack Cursor plugin. |
| Claude Code | untested | SKILL.md format is compatible, but the skill references Cursor paths (.cursor/skills, ~/.cursor/rules) and the Task tool. Untested. |
| Grok Build | untested | Grok reads SKILL.md folders; Cursor-specific paths would need rewriting. Untested. |
| Codex | untested | Untested. |
| AGENTS.md generic | untested | Untested. |
By model family
| Claude | default reviewer/worker model (claude-opus-5-5) via setup-pstack |
| GPT | untested |
| Grok | default second model (grok-4.7) via setup-pstack |
| Gemini | untested |
Sponsor slot: your verification tool here, one sponsor per week. Details
Proof of the week
One verification pattern every week, adapted per harness and model. No spam.