The Agent Skills Directory

[COMMAND_EXECUTION]: The Python CLI implements a background worker architecture for long-running 'pro' and 'ultra' analysis modes by spawning the CLI itself as a background process using subprocess.Popen in scripts/cli/worker_utils.py.\n- [COMMAND_EXECUTION]: The notification utility in scripts/cli/notify.py executes agent runtime platform tools (openclaw, clawdbot) via subprocess.run to relay asynchronous status updates back to the user session.\n- [REMOTE_CODE_EXECUTION]: A build-from-source mechanism in alibabacloud-data-agent-mcp-skill/scripts/select-binary.sh uses go build to compile the Go MCP server from the included source code if a pre-compiled binary is not available for the host platform.\n- [PROMPT_INJECTION]: The skill handles external data from databases and user-uploaded files (CSV, XLSX, JSON, TXT) for analysis, presenting an inherent surface for indirect prompt injection.\n
Ingestion points: Local files uploaded through data_agent_upload_file and database tables managed in the Alibaba Cloud Data Center.\n
Boundary markers: None explicitly defined; data is analyzed directly by the AI agent based on natural language queries.\n
Capability inventory: The skill can execute subprocesses for task management and notifications, and has write access to its local sessions/ directory for logging and persistence.\n
Sanitization: The skill relies on the host AI model's built-in safety guidelines and guardrails during the analysis of external data content.

alibabacloud-data-agent-skill