user-cohort-forensics
Pass
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill documents the intended use of Amplitude-specific tools for data analysis. The workflows described (resolving identities, sampling populations, and auditing events) are standard functionalities for the Amplitude platform.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external identifiers such as emails, user IDs, and cohort names, which constitutes an ingestion point for data that could theoretically contain malicious instructions.
- Ingestion points: Processes user-supplied emails and identifiers for identity resolution in
SKILL.md. - Boundary markers: No explicit delimiters or instructions to ignore embedded content are present in the provided markdown.
- Capability inventory: Utilizes data retrieval tools including
get_amp_user_data,query_amplitude_data, anduse_amplitude_cohortsto fetch user profiles and timelines. - Sanitization: Relies on the underlying platform's tool-level validation and agent guardrails to manage the processing of external data.
Audit Metadata