create-agentsmd-instructions
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to read untrusted codebase files and incorporate their content into a newly generated instructions file.
- Ingestion points: Project codebase files and existing configuration files such as
AGENTS.md,.github/copilot-instructions.md, andCLAUDE.md. - Boundary markers: Absent. There are no instructions for the agent to use delimiters or to disregard instructions embedded within the analyzed project files.
- Capability inventory: File system write operations to create or update the
AGENTS.mdfile. - Sanitization: Absent. The skill does not specify any validation or sanitization of the data extracted from the codebase before it is written.
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to review documentation from well-known and established services.
- Evidence: References official documentation from GitHub Copilot and technical guidelines from the Anthropic (Claude) website to inform the creation of the file.
Audit Metadata