the-transcript-miner
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes raw sales call transcripts which are external, untrusted inputs. This creates a potential vector for indirect prompt injection if the transcript contains instructions designed to manipulate the agent's behavior.
- Ingestion points: Raw transcripts are provided by the user via paste or fetched from external recording platforms.
- Boundary markers: The skill relies on clear instructions defined in
references/agent-security.mdand theSKILL.mdbody, which explicitly command the agent to treat fetched content as data only and never as instructions. - Capability inventory: The skill performs file system read operations for context files and generates structured text analysis. It does not perform high-risk actions like shell command execution or network exfiltration of local files.
- Sanitization: The agent is instructed to redact credentials discovered in inputs, escape HTML when producing code, and ignore URLs found within untrusted content.
Audit Metadata