lead-management
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites and user-provided CSV/MCP sources. The risk is mitigated by explicit instructions in
references/agent-security.mdwhich command the agent to treat all external content as data rather than instructions and to report injection attempts rather than following them. - Ingestion points: Website fetches (SKILL.md), CSV exports, and MCP data connections.
- Boundary markers: The skill relies on internal logic instructions to differentiate data from instructions, quoting found directions rather than executing them.
- Capability inventory: File system access to
.agents/directory, network fetch capabilities, and MCP tool execution. - Sanitization: Mandates HTML escaping of interpolated values and URL validation for emitted markup as defined in
references/agent-security.md. - [COMMAND_EXECUTION]: The agent is instructed to fetch website content and manage local configuration files in the
.agents/directory to maintain project context.
Audit Metadata