dotnet-agent-instructions
Pass
Audited by Gen Agent Trust Hub on Nov 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a bundled Python script (
scripts/install-agent-instructions.py) to automate the installation of instruction blocks. The script uses standard libraries (pathlib,argparse) and performs localized file reading and writing within the target repository. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to modify instruction files (
AGENTS.md,CLAUDE.md) that influence agent behavior in future sessions. - Ingestion points: Reads the contents of existing
AGENTS.mdorCLAUDE.mdfiles in the repository root. - Boundary markers: Uses explicit HTML comment markers (
<!-- dotnet-skills:self-review:start -->) to delimit the injected instruction block. - Capability inventory: The skill has the capability to write to the local file system (
Path.write_text). - Sanitization: The script performs structural validation to ensure only one instruction block exists and that markers are matched, but it does not sanitize the existing content of the files it modifies.
Audit Metadata