The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill includes a maintainer script, scripts/fetch_workflow_examples.py, which utilizes the scenario-sdk Python package. This is the official SDK for the vendor's API. \n- [INDIRECT_PROMPT_INJECTION]: The skill performs ingestion of workflow graph data from external sources, such as public templates or exports from other tools (ComfyUI, Weavy). This represents a potential indirect injection surface. \n
Ingestion points: Retrieval of workflow data via the workflow_get tool and the processing of third-party JSON exports as described in references/foreign-graph-import.md. \n
Boundary markers: The instructions do not specify explicit delimiters for untrusted graph data. \n
Capability inventory: The agent uses tools like workflow_create, workflow_update, and workflow_publish to apply changes based on the ingested content. \n
Sanitization: The skill documentation provides specific guidance to reduce external files into summarized tables before processing, which acts as a mitigation against injection and context-window exhaustion. \n- [COMMAND_EXECUTION]: The documentation contains setup instructions for users to install sibling skills using the npx skills add command. This is a standard platform-level operation and does not involve the automated execution of untrusted input.

scenario-workflow-authoring