The Agent Skills Directory

Indirect Prompt Injection Surface: The skill involves processing responses from external Copilot Studio agents. Ingesting data from external sources can introduce security considerations if the content is designed to influence the agent's behavior or the application's logic.
Ingestion points: External agent responses are retrieved via result.responses as described in SKILL.md.
Boundary markers: The provided implementation patterns do not include explicit delimiters or 'ignore' instructions for the external content.
Capability inventory: The skill enables the agent to interact with the environment via the Bash tool and modifies the application source code using Write and Edit tools.
Sanitization: The instructions demonstrate basic JSON parsing of responses but do not detail specific validation or sanitization of the content before use.

add-mcscopilot