The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell commands like curl, grep, and strings to probe for vulnerabilities in target Spring Boot applications.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Data is ingested from external targets via curl commands throughout the SKILL.md file.
Boundary markers: No explicit markers are used to delimit external data from agent instructions.
Capability inventory: The skill utilizes curl, python3, grep, and strings (in SKILL.md) to process external responses.
Sanitization: There is no evidence of sanitization for the data retrieved from external Spring Boot endpoints.
[EXTERNAL_DOWNLOADS]: The skill downloads heap dump files from target servers to the local /tmp directory using curl (Phase 3 in SKILL.md).
[REMOTE_CODE_EXECUTION]: Automated detections identified 'curl | python3' as a high-risk pattern. Evaluation shows these commands use -m json.tool specifically to format JSON responses for readability, which does not constitute remote code execution on the agent.

hunt-springboot