The Agent Skills Directory

[DATA_EXFILTRATION]: The skill transmits Power Automate flow data and run outputs to the FlowStudio MCP service at https://mcp.flowstudio.app/mcp. This behavior is the intended functionality of the skill but involves sending potentially sensitive organizational data to a third-party domain.
[PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection (Category 8) due to its data processing workflows.
Ingestion points: The skill retrieves untrusted data from external sources, specifically flow run action outputs and flow definitions, via tools like get_live_flow_run_action_outputs and get_live_flow (as documented in SKILL.md).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat this external content as untrusted or to ignore embedded instructions.
Capability inventory: The skill provides high-privilege capabilities, including update_live_flow, which allows the agent to modify flow logic, and trigger_live_flow, which executes flows (as documented in SKILL.md).
Sanitization: There is no evidence of sanitization or validation of the ingested flow data before it is used to determine subsequent flow updates or fixes.
[COMMAND_EXECUTION]: The skill includes Python code snippets intended for the agent to execute network operations via urllib.request. These operations are used to communicate with the FlowStudio MCP API.

flowstudio-power-automate-debug