api-fuzzing-bug-bounty
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to perform reconnaissance and data extraction tasks using external command-line tools such as Kiterunner (
kr) and Python-based utility scripts (json2paths.py) on the local system. - [EXTERNAL_DOWNLOADS]: Provides a comprehensive list of external tools, wordlists, and security repositories hosted on third-party platforms including GitHub, Gist, and Netlify for the purpose of security testing.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data in the form of API documentation and server responses, creating a potential surface for indirect prompt injection.
- Ingestion points: The skill ingests
swagger.json,openapi.json, and the content of live API responses from target systems. - Boundary markers: Lacks technical delimiters or instructions to ignore embedded commands within ingested data, though it incorporates a manual confirmation gate requiring user approval for high-risk actions.
- Capability inventory: The skill utilizes capabilities for executing shell commands, running scripts, and performing network requests via
curl. - Sanitization: No specific instructions are provided for sanitizing or validating the contents of external API schemas or responses before they are analyzed or processed by the agent.
Audit Metadata