voice-extractor
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by extracting stylometric features and idioms from untrusted user writing samples and incorporating them into prompt templates for other agents. This could allow an attacker to embed instructions within samples that are then executed by downstream models.
- Ingestion points: Writing samples are ingested from user-provided text, files, or folders during the voice extraction process.
- Boundary markers: The skill wraps injected content in
<voice_fingerprint>delimiters to distinguish it from static instructions. - Capability inventory: The skill is capable of reading and writing YAML configuration files in the local
~/.newsjack/voice/directory. - Sanitization: No explicit sanitization or validation of the ingested sample content is specified to prevent the extraction and subsequent injection of malicious prompt strings.
Audit Metadata