tech-writer
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The instructions use high-pressure emotional language ('If you not perform well enough YOU will be KILLED') to mandate high-quality output. While intended for performance reinforcement, this aligns with patterns used to override AI constraints.
- [COMMAND_EXECUTION]: The skill provides the agent with shell commands such as
findandgrepto perform codebase analysis and locate documentation files on the host system. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes content from the local codebase and external documentation sources while maintaining execution capabilities.
- Ingestion points: File discovery and content processing steps in
SKILL.md(e.g., Phase 1: Assessment and Planning). - Boundary markers: Absent; the skill does not instruct the agent to ignore or delimit instructions found within the analyzed files.
- Capability inventory: Shell command execution (
find,grep) and integration with theContext7 MCPtool. - Sanitization: No content sanitization or validation routines are specified for the data being processed.
Audit Metadata