The Agent Skills Directory

[SAFE]: The skill operates as intended for tuning Monte Carlo monitors and does not exhibit malicious patterns. It follows best practices by requiring user confirmation for all modifications.
[DATA_EXFILTRATION]: The skill fetches monitor data and alert history from the authenticated monte-carlo-mcp server. This data is used solely for analysis within the agent's context. No evidence of data being sent to unauthorized external domains was found.
[COMMAND_EXECUTION]: The skill uses specific MCP tools for its core logic. It does not invoke arbitrary shell commands or access system-level resources outside of the MCP environment.
[PROMPT_INJECTION]: The skill processes data from the Monte Carlo API, which could potentially contain untrusted content, creating a surface for indirect prompt injection. However, the human-in-the-loop requirement for applying changes effectively mitigates this risk.
Ingestion points: get_monitor_report and get_monitors tool outputs in SKILL.md.
Boundary markers: No explicit delimiters are used when processing the fetched report data.
Capability inventory: Monitor update tools (create_or_update_*_monitor, tune_*_table_monitor) as defined in Phase 5.
Sanitization: The skill does not perform explicit sanitization of the incident data before analysis.
[REMOTE_CODE_EXECUTION]: No remote code execution, external package installation, or dynamic script generation was detected. The skill relies on local reference files and pre-configured MCP tools.

tune-monitor