The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several Azure CLI (az) commands to check account status, perform login, and retrieve access tokens for authentication.
Evidence: Steps 2a.1 and 2a.2 detail the use of az account show, az login, and az account get-access-token.
[DATA_EXFILTRATION]: While the skill retrieves an Azure access token, it is used solely to authenticate requests to the official Power Apps API (api.powerapps.com) to list environments.
Evidence: Step 2a.3 describes calling GET https://api.powerapps.com/providers/Microsoft.PowerApps/environments with the Bearer token.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (Azure API responses and user arguments) which could theoretically contain malicious instructions.
Ingestion points: Data is ingested from the Azure CLI output, the Power Apps API JSON response, and the $ARGUMENTS variable.
Boundary markers: No specific delimiters or instructions are provided to the agent to ignore embedded commands in the API response.
Capability inventory: The skill can read/write local files (MCP config) and execute shell commands (az, mkdir).
Sanitization: The skill performs basic sanitization such as stripping trailing slashes and extracting subdomains for server names.

mcp-configure