api-recon
Pass
Audited by Gen Agent Trust Hub on Nov 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill harvests substantial amounts of untrusted data, including JavaScript bundles, HTML, and API responses, from target websites for analysis. This creates a vulnerability surface where malicious instructions embedded in a compromised or malicious target site could attempt to influence the agent's behavior during the analysis phase.
- Ingestion points: The
harvest_static.pyscript downloads target JS/HTML files, andruntime_harvest.jsintercepts live API responses. - Boundary markers: The instructions do not explicitly mandate strict boundary markers or warn the agent to ignore potentially malicious instructions embedded within the harvested data.
- Capability inventory: The skill has the capability to execute Python and Node.js scripts, write to the local file system, and perform network operations.
- Sanitization: While the skill extracts specific patterns (APIs, routes, parameters), the underlying content is not formally sanitized before being presented to the agent for higher-level analysis.
- [DYNAMIC_EXECUTION]: The operational workflow requires the agent to copy reference scripts (Python and Node.js templates) into a local directory, modify them to adapt to the specific target application, and then execute them. This involves the dynamic adaptation and execution of code.
- [COMMAND_EXECUTION]: The skill's core functionality is implemented through the execution of external scripts and tools, specifically Python 3 and Node.js (Puppeteer), to perform reconnaissance tasks such as bundle harvesting and dynamic API interception.
Audit Metadata