The Agent Skills Directory

[SAFE]: The skill integrates with established and reputable third-party services including Hunter.io, Apollo.io, SerpAPI, and Groq for data enrichment and processing.
[INDIRECT_PROMPT_INJECTION]: The workflow contains a surface for indirect prompt injection within the 'LLM Extract' node. This node processes search results from SerpAPI using a large language model. If an attacker can influence those search results (e.g., via a compromised or specifically crafted LinkedIn profile), they could potentially influence the extraction logic.
Ingestion points: Webhook node in references/waterfall-template.json which accepts user-supplied lead data.
Boundary markers: The LLM Extract node utilizes a system prompt to constrain the model's behavior to extraction tasks.
Capability inventory: The template includes capabilities for network requests (HTTP Request nodes), database operations (MySQL nodes), and AI-driven data processing.
Sanitization: Basic string normalization (trimming, case conversion) is performed in the 'Normalize Input' node, though it does not specifically filter for malicious prompt instructions.
[COMMAND_EXECUTION]: The Cache Insert node in the provided n8n template constructs SQL queries using string interpolation and manual escaping (replacing single quotes). While this is a common implementation pattern within the n8n ecosystem, it represents a manual approach to database security.

enrichment-waterfall