client-proposal-generator
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to perform web research on client companies using the WebSearch tool to personalize proposals. This creates a surface where malicious content from external websites could be ingested and influence the agent's behavior or output content.
- Ingestion points: The research process in 'references/research.md' uses WebSearch to pull company overviews, news, and technology signals.
- Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded prompts within the gathered research data.
- Capability inventory: The agent has access to Write, Bash, WebSearch, and Glob tools.
- Sanitization: There is a quality check for accuracy, but no explicit sanitization or filtering logic is provided to handle potentially malicious external content.
- [COMMAND_EXECUTION]: The instructions in 'references/research.md' direct the agent to use the Bash tool for gathering technology signals, which might include inspecting local files or performing system-level checks. While this is intended for data collection, the use of a shell-based tool requires monitoring to prevent the execution of unintended or unsafe commands.
Audit Metadata