expert-graham
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is composed exclusively of markdown files (
SKILL.mdand several reference files) that provide context and frameworks for startup analysis. There is no active code (scripts, binaries, or configuration) that performs network operations, file modifications, or sensitive data access. - [EXTERNAL_DOWNLOADS]: The skill includes instructions to use the
openskillsCLI tool vianpxto load companion modules (expert-engine,expert-graham). This is a standard instructional pattern for the OpenSkills framework and does not represent a security risk within the skill's own implementation. - [PROMPT_INJECTION]: The skill provides stylistic guidance and conversational framing (e.g., in
judgment-filters.md) to help the agent emulate Paul Graham's thinking style. These instructions are focused on providing higher-quality startup advice and do not attempt to bypass safety filters or override system constraints.
Audit Metadata