dotnet-agent-instructions
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run a local Python script (
install-agent-instructions.py) to automate the installation of instructions. This script is bundled with the skill and operates locally on project files. - [INDIRECT_PROMPT_INJECTION]: The skill modifies agent instruction files (
AGENTS.mdorCLAUDE.md) which influences the agent's behavior in subsequent turns. - Ingestion points: The script reads and writes to
AGENTS.mdandCLAUDE.mdin the repository root. - Boundary markers: The script explicitly uses HTML comments (
<!-- dotnet-skills:self-review:start -->and<!-- dotnet-skills:self-review:end -->) to delimit the injected instructions. - Capability inventory: The script performs file read and write operations using Python's
pathlib. It does not perform network operations or access sensitive system paths. - Sanitization: The script includes logic to detect unmatched markers or multiple blocks, refusing to update if the file structure is inconsistent.
- [DYNAMIC_EXECUTION]: The skill generates markdown content programmatically and writes it to the file system. The content being written is a static template defined within the script.
Audit Metadata