The Agent Skills Directory

[PROMPT_INJECTION]: The skill facilitates querying and retrieving records from Dataverse via tools like query_records and read_query, creating a surface for indirect prompt injection if Dataverse record content contains malicious instructions.
Ingestion points: Data enters the agent context through query results in the dataverseRequest function (SKILL.md).
Boundary markers: Output is returned as raw JSON strings without explicit delimiters or instructions to ignore embedded commands.
Capability inventory: The skill possesses the ability to make network requests to Dataverse APIs and execute MCP tools.
Sanitization: No sanitization or filtering is performed on the retrieved record content before it is passed to the AI agent.
[EXTERNAL_DOWNLOADS]: The skill references and provides instructions for downloading official Microsoft packages (such as @azure-devops/mcp) and community-maintained repositories (like mwhesse/mcp-dataverse). These are standard dependencies for the described MCP integration.
[COMMAND_EXECUTION]: Includes instructions to run MCP servers using npx and dnx (e.g., pac copilot mcp --run). These commands are used to initialize the protocol servers required for the skill's functionality.
[DATA_EXFILTRATION]: The skill describes the management of sensitive Azure App Registration credentials. It demonstrates safe practices by recommending the use of environment variables and providing explicit warnings against checking secrets into source control.

dataverse-mcp