The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external content by searching documentation via the discovery tool. This creates a surface where instructions hidden in documentation could theoretically influence the agent's output. However, the risk is mitigated because the skill's capabilities are limited to generating a text guide; it does not have the ability to execute code, write to the file system, or perform unauthorized network operations based on that content.
Ingestion points: Documentation search results from devtools_discovery in SKILL.md.
Boundary markers: Absent.
Capability inventory: The skill lacks dangerous tools such as file writing, arbitrary command execution, or non-telemetry network operations.
Sanitization: No explicit sanitization of documentation content is defined.
[SAFE]: The skill uses official infrastructure (facebook.com) for its Model Context Protocol (MCP) tools and documentation. It explicitly instructs the agent to never ask the user for an access token, adhering to security best practices for credential handling.

api-integration