aif-improve
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a restricted Bash tool to execute git commands for branch discovery and file status checks, which are used solely to locate and manage implementation plan files.
- [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes data from the project codebase and implementation plans to generate suggestions. 1. Ingestion points: Reads from .ai-factory/PLAN.md, research files, and source code files identified via search tools. 2. Boundary markers: Content is ingested without explicit delimiters or instructions to ignore embedded commands. 3. Capability inventory: The skill can modify implementation plans using Write and Edit and manage task lists via TaskUpdate. 4. Sanitization: No specific validation or escaping is applied to the ingested text during analysis.
Audit Metadata