Unverified50% confidenceOpinionExact time
MAQS-Way author Aleksandr Kossarev comes from Jõgeva, Estonia, and his project stance is: "Don't check whether a mechanism exists, check whether it actually works."
1
Sources
50%
Confidence
Long-term
Relevance
10/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedVerification is indispensable in agent systems; a model claiming that tests have passed does not prove the tests actually passed, and the harness should collect evidence from actual test execution and report it faithfully.59% similarVerifiedAgent tests should programmatically inspect the final state to verify the true end state of the system rather than relying on the model's self-report.58% similarUnverifiedThe author notes that the tool is essentially pattern matching rather than proof, meaning a scan hit only indicates the need for manual review, and a clean scan result does not mean the project is absolutely secure.56% similarUnverifiedThe project explicitly identifies where the benchmark no longer matches real-world workloads.56% similarUnverifiedDeveloper olgazju built a Kaggle benchmark called "Where Did My Data Go?" to test whether LLMs honestly admit they cannot determine the data's whereabouts when evidence is insufficient.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1045072API
curl https://kongchang.com/api/v1/knowledge/claims/1045072MCP
get_claim(id=1045072)