The Agent Skills Directory

[SAFE]: The skill is a technical documentation suite authored for development and deployment guidance. It does not contain malicious code, instructions, or obfuscated patterns.
[EXTERNAL_DOWNLOADS]: References official, well-known libraries and services including Microsoft's Agent Framework, CopilotKit, and Azure Foundry. Dependencies identified (e.g., @copilotkit/runtime, agent-framework-foundry) are standard packages from trusted ecosystems.
[COMMAND_EXECUTION]: Provides instructions for using standard development tools such as the Azure Developer CLI (azd), curl for API testing, and pip for dependency management. All commands are contextually appropriate for the stated purpose of application development and debugging.
[CREDENTIALS_UNSAFE]: Adheres to security best practices by recommending the use of environment variables (e.g., FOUNDRY_PROJECT_ENDPOINT) and managed identities (Entra auth) with specific scopes. It explicitly warns against exposing sensitive credentials or backend endpoints to the frontend/client-side.
[DATA_EXFILTRATION]: No unauthorized data collection or exfiltration patterns were detected. Network operations are directed towards legitimate, well-known cloud services (Azure, GitHub, Microsoft).
[PROMPT_INJECTION]: The skill identifies security-relevant platform behaviors, such as the 'duplicate-execution hazard' in tool calls, and provides specific programmatic mitigations to prevent accidental or malicious re-execution of sensitive actions.

foundry-hosted-agent-copilotkit