The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external OAuth 2.0 discovery endpoints to perform its auditing functions.
Ingestion points: scripts/agent.py and scripts/process.py fetch JSON metadata from user-provided issuer URLs.
Boundary markers: The scripts do not implement specific boundary markers in their output for the external data.
Capability inventory: The skill performs network GET requests and prints findings to the console; it does not execute code or modify files based on the ingested data.
Sanitization: The scripts use standard JSON parsing but do not perform additional sanitization on metadata fields before display. This represents a minor surface where a malicious server could provide deceptive text, but given the skill's purpose as an auditor, this is expected behavior.
[COMMAND_EXECUTION]: The skill provides scripts (agent.py and process.py) intended for local execution to perform security assessments. These scripts do not execute arbitrary shell commands based on external inputs and focus solely on the intended auditing logic.

configuring-oauth2-authorization-flow