The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell scripts to perform active security testing. It executes standard utilities including curl, awk, jq, grep, and sed to interact with web endpoints and process response data.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from external web servers (HTTP headers and response bodies). This creates a vulnerability surface where a malicious target application could return crafted content intended to influence the agent's reasoning or the outcome of the security audit.
Ingestion points: Captured HTTP responses from the user-defined TARGET using curl across various phases in SKILL.md.
Boundary markers: Absent; the skill does not use specific delimiters or instructions to treat the captured network data as strictly untrusted text for the LLM.
Capability inventory: The skill possesses network communication capabilities (curl) and local file management (mktemp, rm) within the bash environment.
Sanitization: Relies on jq for parsing JSON payloads and awk for cookie extraction, which provides structural filtering but not content sanitization against prompt injection patterns.

hunt-session