The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze data from untrusted external sources, such as HTTP response bodies, API documentation (Swagger/OpenAPI), and JavaScript files from target web applications. This creates a surface for indirect prompt injection if those sources contain malicious instructions.
Ingestion points: SKILL.md (Phase 1 authentication discovery and Phase 2 Mass Assignment analysis) and references/web-privesc-techniques.md (JavaScript extraction and response tampering).
Boundary markers: The instructions do not specify the use of delimiters or specific warnings to ignore instructions embedded within the target data.
Capability inventory: The skill methodology requires the agent to perform network requests (GET, POST, PUT, PATCH, DELETE), manipulate HTTP headers, and execute Python scripts for cookie decoding and session forgery.
Sanitization: There is no evidence of sanitization or validation logic for the external content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill methodology involves instructions for the agent to execute shell commands and scripts to perform security testing tasks. This includes using tools like urlfind for URL extraction and running Python snippets for Base64 manipulation and Flask session signing as described in references/web-privesc-techniques.md.

privilege-escalation-web