dedupe-contacts
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a diagnostic command
clay whoamito verify the workspace configuration and ensure necessary tools are present before proceeding with data processing. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from contact lists, which presents a surface for indirect prompt injection.
- Ingestion points: Contact lists supplied via CSV files, Clay tables, or CRM exports (SKILL.md).
- Boundary markers: The instructions do not define specific delimiters for separating data from instructions, but they mandate the use of deterministic code-based normalization rather than relying on an LLM for initial processing.
- Capability inventory: The skill is limited to reading data, performing environment checks, and outputting a text-based merge plan. It is explicitly prohibited from executing merges, deleting records, or writing to CRM systems.
- Sanitization: The skill incorporates normalization routines including case folding, trimming, suffix stripping, and apex domain normalization to clean inputs before comparison.
Audit Metadata