The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of Node-RED flows designed to ingest data from untrusted external sources, such as MQTT brokers and HTTP REST endpoints. These patterns introduce a surface for indirect injection attacks if processed data is not strictly validated before being used in downstream logic or storage operations.
Ingestion points: http in nodes in assets/templates/http_api_flow.json and mqtt in nodes in assets/templates/mqtt_flow.json.
Boundary markers: SKILL.md includes a validation checklist that advises verifying JSON syntax and wire connections, though it does not explicitly mandate sanitization of incoming message payloads.
Capability inventory: The Node-RED environment supports function nodes for JavaScript execution (documented in references/function_snippets.md) and exec nodes for system command execution (documented in references/node_schemas.md).
Sanitization: The provided templates include basic authentication checks and status code responses, but comprehensive input sanitization remains the responsibility of the developer.
[DYNAMIC_EXECUTION]: A core component of the skill is providing templates and boilerplate code for Node-RED function nodes (assets/boilerplate/function_async.js). These nodes execute arbitrary JavaScript code within the Node-RED runtime. While this is the intended purpose of the skill, it represents a dynamic execution vector that requires careful oversight to prevent the execution of malicious logic embedded in flow configurations.
[CREDENTIALS_UNSAFE]: The skill's documentation and templates include hardcoded placeholder credentials. Specifically, references/api_reference.md contains a curl example with admin:password, and assets/templates/http_api_flow.json includes a demo API key demo-api-key-123 in its authentication logic. These should be treated as placeholders and replaced by users with secure, environment-specific secrets.

node-red