ai-dev-tools-sync
Pass
Audited by Gen Agent Trust Hub on Mar 7, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes instructions from external configuration files and web search results, which introduces a surface for indirect prompt injection.
- Ingestion points: Reads project-level configuration files including
CLAUDE.md,.github/common.copilot-instructions.md, and.github/workspace.copilot-instructions.md, as well as web search results for tool features. - Boundary markers: No explicit delimiters are used to separate untrusted file content from the agent's primary instructions.
- Capability inventory: The skill utilizes task planning (
TaskCreate), file reading/writing, and web search capabilities. - Sanitization: No automated sanitization is performed, though the skill explicitly instructs the agent to 'Be skeptical' and require 'traced proof' for every claim.
- [EXTERNAL_DOWNLOADS]: The skill references official documentation and resources from well-known services and trusted organizations.
- Evidence: Links to
docs.github.com,code.visualstudio.com, and the official GitHubawesome-copilotrepository are provided for feature research.
Audit Metadata