CacheDeception
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from HTTP requests and responses via the Caido MCP.
- Ingestion points: Web traffic history is retrieved through
mcp__caido__list_by_httpqlas shown inSKILL.md. - Boundary markers: The instructions lack explicit boundary markers or directives for the agent to ignore potentially malicious content within the analyzed HTTP responses.
- Capability inventory: The skill has the capability to send HTTP requests (
mcp__caido__sendRequest,mcp__caido__start_replay_task), open browser tabs, and create findings in an external application (mcp__caido__create_findings_from_requests). - Sanitization: There is no evidence of sanitization, validation, or filtering of external content before it is processed by the agent or interpolated into findings/reports.
- [COMMAND_EXECUTION]: The skill provides templates and JavaScript helper functions in
SKILL.mdto generate complex exploit payloads and command-line instructions for tools likeffuf. While these are intended for security testing, they involve generating and executing network-active commands based on user-supplied endpoints and observed traffic.
Audit Metadata