build-talk-outline
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses
~/.devadvokit.mdand../../shared/ai-antipatterns.mdto gather speaker context and perform quality checks. These are local file reads necessary for the skill's documented functionality and do not involve any network exfiltration or sensitive data exposure. - [COMMAND_EXECUTION]: No shell commands, dynamic context injections, or suspicious scripts were detected in the skill files.
- [PROMPT_INJECTION]: The skill uses instructional framing to ensure the agent follows specific presentation frameworks (Winston principles). These instructions are focused on content quality and do not attempt to bypass safety guidelines.
- [PROMPT_INJECTION]: The skill processes user-provided talk abstracts and Q&A answers. While this represents an ingestion surface for potential indirect prompt injection, the skill lacks the capabilities (such as network access or command execution) to be exploited via this vector.
Audit Metadata