Unverified50% confidenceDecision RuleExact time
Test drift makes scores incomparable, so you must lock the task set version and prompt version, and record the model ID rather than "latest version".
1
Sources
50%
Confidence
Long-term
Relevance
10/15/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen debugging production Agent failures, you should log the model version and temperature parameter of each call, because model providers silently update models, and without version logs there is no way to correlate model changes with failures.71% similarUnverifiedThe prompt maintenance process for a new model version includes: reading the release notes, running without modifying the prompt to look for regressions, changing only one variable at a time, and keeping the old version until the new prompt is confirmed to perform better.69% similarUnverifiedMitigating prompt drift requires pinning the LLM version and parameters, recording the full prompt and response for every step, and running regression tests; but silent updates from model vendors and changes in tool return formats also cause drift, so prompts, model versions, parameters, and tool schemas should be versioned together as one set of immutable configuration.67% similarUnverifiedWhen a tool schema changes but the corresponding instruction descriptions in the prompt are not updated in sync, the model may generate malformed tool call requests, leading to runtime exceptions.67% similarUnverifiedModel parameters from different generations or base versions are not interchangeable, and the complete version number must be locked down when comparing results.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1037047API
curl https://kongchang.com/api/v1/knowledge/claims/1037047MCP
get_claim(id=1037047)