echo-segment
Warn
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructs the agent to scan the local codebase for data sources, explicitly identifying "support tool configs" and "user records" as targets. This leads the agent to access potentially sensitive files containing credentials, API keys, or personally identifiable information (PII).
- [PROMPT_INJECTION]: The skill processes untrusted external data such as support tickets, NPS verbatims, and app reviews which may contain malicious instructions.
- Ingestion points: Ingests external text from customer support tickets and public reviews as described in Step 1.
- Boundary markers: No markers or instructions are provided to the agent to treat this data as untrusted or to ignore embedded commands.
- Capability inventory: The agent has access to filesystem tools (Read, Grep, Glob) and shell access (Bash), which can be abused if an injection is successful.
- Sanitization: No sanitization or validation of the external content is performed before processing.
Audit Metadata