The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell-based commands (specifically curl) to communicate with the n8n REST API. These commands are used for standard operational tasks such as creating, updating, activating, and deleting workflows, as well as inspecting execution logs.
[EXTERNAL_DOWNLOADS]: The documentation references the n8n-mcp server (available at github.com/czlonkowski/n8n-mcp), which is the recognized community standard for connecting n8n instances to Model Context Protocol clients. The skill provides instructions for running this tool via npx or Docker to facilitate agent-based workflow management.
[PROMPT_INJECTION]: As the skill is designed to ingest and execute workflow designs in JSON format, it possesses a surface for indirect prompt injection if the source of those designs is not properly vetted. The skill addresses this by providing specific validation and sanitization steps.
Ingestion points: Workflow JSON payloads processed by the POST /workflows endpoint or provided as arguments to the n8n_create_workflow MCP tool.
Boundary markers: The instructions explicitly recommend performing validation (using n8n_validate_workflow) prior to any creation or update operation.
Capability inventory: The skill facilitates full CRUD (Create, Read, Update, Delete) operations on workflows and management of execution data through the API and MCP tools.
Sanitization: The skill identifies n8n_autofix_workflow as a mechanism to repair and sanitize structural errors within workflow JSON before deployment.
[SAFE]: The skill demonstrates a strong security posture by advising against hardcoding credentials, encouraging the use of least-privilege API scopes, and recommending the DISABLED_TOOLS environment variable to prevent unauthorized destructive actions in agent-facing environments.

n8n