Unverified50% confidenceFactExact time
In agent-audit-mcp, decision thresholds (such as what counts as high risk and what counts as missing evidence) can be overridden via environment variables.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
Unverifiedagent-audit-mcp's audit decision rules include: a hit on v6 context permission overreach is judged BLOCK; a hit on v7 missing causal evidence or v8 high value risk is judged ALERT; a Hamming distance > 4 between adjacent steps is judged WATCH; if none is triggered, it is OK.77% similarUnverifiedOnce MCP tools are exposed, an Agent's invocation decisions depend on the model's judgment, and in production environments permission checks, parameter whitelists, and audit logs still need to be added at the outer layer.75% similarUnverifiedExplainability should be designed according to risk classification: basic post-hoc explanation suffices for low-risk scenarios, data lineage tracking is required for medium risk, and the highest risk requires linking explanations to specific individuals.72% similarUnverifiedA reasonable approach is to tier by risk: read-only, reversible, and low-impact actions can trust upstream checks, while actions that produce real side effects, are irreversible, or involve sensitive resources require independent verification at the execution boundary.68% similarUnverifiedThe key variables in comparing risk are not open source or closed source itself, but the attacker's capability, how the model is deployed, and whether there is accompanying abuse monitoring.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1057189API
curl https://kongchang.com/api/v1/knowledge/claims/1057189MCP
get_claim(id=1057189)