The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of security-focused packages and tools, such as o365spray and the ProjectDiscovery toolkit, via standard package managers like pip and go install from well-known repositories.
[COMMAND_EXECUTION]: Provides instructions for running reconnaissance tools (msftrecon) and executing custom Python scripts that automate credential validation and browser-based authentication flows using Playwright.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external credential lists and email addresses, creating an ingestion surface for potentially untrusted data.
Ingestion points: email and password variables processed in the attempt() and saml_validate() functions within SKILL.md.
Boundary markers: Absent; the skill does not implement specific delimiters or 'ignore' instructions for the external input data.
Capability inventory: Performs local file writes for logging (json.dump), network requests (urllib.request), and browser automation (playwright).
Sanitization: Implements standard URL encoding via urllib.parse.urlencode to sanitize inputs before they are included in ROPC request bodies.

m365-entra-attack