skills.rest
allemaar
allemaarCOMMUNITY
Explore all skills by this author →
Website𝕏@allemaarGitHub

double-check

Re-verifies a specific claim, output, or decision through adversarial multi-pass re-reading of sources.

14|1|Updated Jun 14, 2026
Search Tags:#quality assurance#fact-checking#verification#plan review#claim validation#adversarial review
One-click install
npx skills add https://github.com/allemaar/open-skills --skill double-check-allemaar
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: double-check
Source: https://github.com/allemaar/open-skills/tree/main/skills/double-check
Command: npx skills add https://github.com/allemaar/open-skills --skill double-check-allemaar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When an agent produces a plan, output, or decision, a single-pass review often rubber-stamps errors. This Skill forces a fresh, adversarial re-read of the actual sources so unsupported claims, wrong assumptions, and gaps are caught before you act on them. ## Core Features & Use Cases - Five-Phase Re-Verification: Scope the target, re-read from actual sources, challenge adversarially, propose improvements, and deliver a verdict. - Explicit Verdicts: Every run ends with SOLID, IMPROVED, or RETHINK, each backed by cited evidence. - Iterative Improvement: When issues are found, the Skill proposes concrete fixes and offers to apply them and run another pass. - Use Case: After an agent drafts a migration plan, run /double-check to re-read the referenced code and docs, challenge each step's ordering and assumptions, and receive a verdict on whether the plan holds up. ## Quick Start Ask the agent to run /double-check on the plan or output you want re-verified, for example by saying "double-check this migration plan against the actual source files".

Frequently Asked Questions about double-check

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify an AI-generated plan before executing it?▼

Run /double-check on the plan. The Skill re-reads the actual source files and docs, challenges each step for gaps, wrong assumptions, and ordering issues, then delivers a SOLID, IMPROVED, or RETHINK verdict with cited evidence.

What is the difference between double-check and a formal verification gate?▼

Double-check is a quick, targeted re-read of one claim or artifact. The /verify skill is a formal phase gate requiring intent, plan, and execution artifacts. Use double-check for fast spot checks and escalate to /verify for full gating.

Can double-check verify claims from memory or prior analysis?▼

No. The Skill explicitly requires going back to actual sources such as files, code, and docs. Verifying from memory or cached understanding is prohibited by its rules.

What happens when double-check finds problems in my output?▼

It delivers an IMPROVED verdict listing each issue, a proposed fix, and the reasoning. It then offers to apply the fixes and run another pass, iterating until the target is solid.

When should I not use double-check?▼

Avoid it when you need a formal multi-artifact phase gate (use /verify) or an advisory multi-perspective critique of an output (use /insight-critique). Double-check targets a single claim or artifact.

Related Skills

Explore More Software Engineering Skills→
AutisticAFAutisticAF
4

macos-app-planner

Generate structured planning documents for new and existing macOS apps.

Community
Advanced
swarm-ai-researchswarm-ai-research
39

session-close

Summarize research work, update memory logs, commit changes, and push to version control.

Official
Intermediate
theamtheam
36

research-devil-advocate

Stress-test Limina research directions with adversarial evidence and baseline audits.

Official
Advanced
xz1220xz1220

autopilot

Generate end-to-end working code from a brief product idea.

Community
Advanced