The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands (curl, grep, shasum, cat) to establish baseline soft-404 behaviors, fingerprint target technology stacks, and perform authentication preflights. These operations are performed against user-supplied targets and variables ($TARGET, $HOSTS, $SESSION_COOKIE) as part of legitimate reconnaissance and security testing workflows.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external host responses to drive its logic. Ingestion points: Response headers and HTML bodies fetched via curl operations in the fingerprinting loops. Boundary markers: The skill uses response status, byte length, and body hashes to distinguish between control responses and meaningful signals. Capability inventory: The skill has the ability to execute shell commands and load additional agent skills via the Skill tool. Sanitization: External input is not directly executed as logic; instead, it is parsed for specific framework markers (e.g., __NEXT_DATA__, laravel_session) which are then mapped to a deterministic set of platform-specific skills.

hunt-dispatch