speak-data-handling
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted user data which could potentially contain malicious instructions intended to influence the agent's behavior.
- Ingestion points: User-provided lesson response text and audio transcripts processed by the system (found in references/implementation-guide.md and SKILL.md).
- Boundary markers: None; the skill lacks specific delimiters or instructions to the LLM to ignore embedded commands within the lesson content.
- Capability inventory: The skill has the ability to perform network requests via curl and the SpeakClient, and write to databases or storage backends (found in SKILL.md and references/implementation-guide.md).
- Sanitization: While the skill includes a PII detection and redaction mechanism using regular expressions, it does not explicitly sanitize or validate the content against prompt injection attempts.
Audit Metadata