Description
doublecheck is an agent skill from the github/awesome-copilot collection that runs a three-layer verification pipeline on AI-generated output. Its goal is not to declare what is true, but to extract every verifiable claim in a response, find sources the user can check independently, and flag anything that looks like a hallucination pattern. It works in two modes. In active mode, invoked without specific text, it stays on for the rest of the conversation and adds an inline verification summary after each substantive response, and the user can ask for the full report on any response or turn the mode off. In one-shot mode it checks a given piece of text on request. Before any substantive response, the skill classifies whether the response contains verifiable claims, and it applies its rules to the claims it finds. Web search is used to locate supporting or contradicting sources, and an adversarial review pass looks for weak points in the claims. The skill is useful when a person relies on an AI answer for facts, and wants a trail to check. Its findings depend on the search results, which can be incomplete, so a claim the skill cannot source is not thereby false, and the user should check the sources that matter. The skill is installed with the skills CLI from the github/awesome-copilot repository. Its skills.sh page showed about 4.2K installs, with Gen Agent Trust Hub and Socket passing and Snyk showing a warning at the time of the scan.…
Highlights
- Three-layer verification of AI output
- Extracts verifiable claims
- Finds supporting or contradicting sources
- Adversarial review for hallucination patterns
- Active and one-shot modes
- Installs with the skills CLI
Ratings & Reviews
No ratings yet
Be the first to rate doublecheck.
People also visit
principle-sequence-verifiable-units
Apply to multi-step work (sweeps, migrations, runs of similar edits)
doublecheck FAQ
What does it do?
Extracts the verifiable claims in AI output, finds sources to check them, and flags likely hallucinations.
What are the two modes?
Active mode stays on for the conversation and adds a summary after each response. One-shot mode checks a given text on request.
Does a claim with no source count as false?
No. The sources depend on search results, which can be incomplete, so the user should check the sources that matter.
How do I install it?
Run npx skills add https://github.com/github/awesome-copilot --skill doublecheck.
What security audits does it have?
skills.sh lists three audits: Gen Agent Trust Hub Pass, Socket Pass and Snyk Warn. A Warn or Fail is worth reading before you install.
What are the best doublecheck alternatives?
Popular alternatives to doublecheck include create-verification-skill, maintain-verification-skill and principle-prove-it-works. See all doublecheck alternatives