The Agent Skills Directory

[SAFE]: The skill interacts with the Progress Observability platform via a remote MCP server at mcp.observability.progress.com. This is a legitimate vendor-owned resource consistent with the skill's purpose.
[CREDENTIALS_UNSAFE]: The skill utilizes an environment variable OBSERVABILITY_MCP_API_KEY for authentication. This follows security best practices by avoiding hardcoded secrets.
[COMMAND_EXECUTION]: A curl command is provided in references/mcp.md as a documentation example for users to manually verify connectivity. This is not an automated or malicious execution pattern.
[SAFE]: The skill documentation in references/mcp.md explicitly addresses the risk of indirect prompt injection from trace content. It provides clear instructions for the agent to treat external content as data rather than instructions, to scrub PII, and to defang boundary tokens (e.g., <input> to < input >).

cost-report