Unverified50% confidenceOpinionExact time
Lauren believes every comment left under a PR saying "this shouldn't be done here" should be treated as a code smell, and the right approach is to turn the constraint into a hard rule, a lint rule, or a CI failure.
1
Sources
50%
Confidence
Long-term
Relevance
10/21/2026
First Seen
Sources
Former Cursor Engineer's 5-Month Experiment: How to Get AI Agents to Auto-Merge PRs
bilibili晓林呐10/18/2026
Related Entities
Related Claims
UnverifiedWhen writing Evals, RFC-style wording such as must/must not/should should be used in the judge prompt to eliminate ambiguity, and questions should be kept atomic (one question asks only one thing).59% similarUnverifiedRelying on prompt constraints to prevent dangerous operations is inherently unreliable, because the model may be jailbroken, make misjudgments, or forget constraints during long conversations.58% similarUnverifiedFor probabilistic systems, rules written in prompts are suggestions, not law; code, by contrast, is unambiguous and executable, either running or erroring out.58% similarUnverifiedSome comments proposed the concept of a "comprehension threshold": if a piece of code cannot be explained clearly by a reviewer within a reasonable time, it should not be merged into the main branch.57% similarUnverified纯写死规则的意图识别方案泛化能力差，用户换一种说法就识别失败57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1081728API
curl https://kongchang.com/api/v1/knowledge/claims/1081728MCP
get_claim(id=1081728)