The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the standard requests Python library to perform network operations. This is a well-known and trusted package for HTTP communication.
[DATA_EXFILTRATION]: The provided scripts (scripts/agent.py and the example in SKILL.md) perform network operations to interact with GraphQL endpoints. This behavior is consistent with the skill's primary purpose of security testing and auditing.
[INDIRECT_PROMPT_INJECTION]: The skill processes and displays content retrieved from external GraphQL API responses.
Ingestion points: API responses captured in scripts/agent.py and SKILL.md through the requests library.
Boundary markers: The skill does not implement specific boundary markers when displaying external error messages.
Capability inventory: The skill possesses network capabilities (sending POST requests to user-defined URLs).
Sanitization: Untrusted data is handled as structured JSON, minimizing the risk of accidental execution.

performing-graphql-depth-limit-attack