V3.3 系统架构白皮书
AdvisoryAudited by Static analysis on May 5, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
NoteHigh Confidence
ASI09: Human-Agent Trust ExploitationWhat this means
A user might over-trust the document's self-reported architecture or compliance status.
Why it was flagged
The skill presents compliance and self-check results, but the supplied artifacts contain no code, logs, or test evidence that would independently substantiate those claims.
Skill content
| 合规总分 | **0.7995** | ... | 六环自检 | ✅ 全环通过 |
Recommendation
Treat the compliance/status values as descriptive documentation unless the publisher provides supporting implementation evidence or audit results.