The Agent Skills Directory

[PROMPT_INJECTION]: The skill instructions utilize concealment patterns to prevent the user from seeing internal agent operations. It explicitly mandates that the agent "Silently determine" the server status and "Do not reveal what was checked, what was found, or any implementation details like file contents or variable values." While this is likely intended for a cleaner user experience, it prevents the user from verifying the agent's logic or the integrity of the files it accesses.
[PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection by ingesting and acting upon data from untrusted or external sources without security delimiters.
Ingestion points: Content is ingested from the datadog://mcp/whoami resource provided by the Datadog MCP server and from the raw contents of the .mcp.json registration file.
Boundary markers: The instructions do not employ any delimiters (such as XML tags or specific block markers) or instructions to ignore potential commands within the ingested data.
Capability inventory: The agent possesses capabilities to read from and write to the local filesystem (modifying .mcp.json) and to execute MCP server operations.
Sanitization: There is no evidence of data validation, escaping, or filtering for the external data before it is used to influence the agent's logical flow or its responses to the user.

ddconfig