c-level-advisor
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection due to its ingestion of local data that could be influenced by external factors. . Ingestion points: The agent reads sensitive company data and historical board decisions from
~/.claude/company-context.mdand~/.claude/decision-log.md. . Boundary markers: No explicit delimiters or 'ignore instructions' warnings are enforced in the file-reading logic to separate data from system prompts. . Capability inventory: The skill has access to 25 Python-based analysis scripts and a multi-agent orchestration protocol. . Sanitization: While an anonymization protocol is implementation for outbound data, input sanitization for the ingested context files is not explicitly defined. - [SAFE]: All detected behaviors, including the processing of company financials and strategic data, are directly associated with the primary purpose of the skill as an executive advisor. The skill includes robust frameworks for data privacy and inter-agent isolation to mitigate common risks.
Audit Metadata