architecture-decision-record
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/adr-manager.pyutility usessubprocess.runto retrieve the current user's name from Git configuration (git config user.name). This execution is safe as it uses a fixed list of arguments, does not invoke a shell, and is used solely for populating the 'Authors' field in generated documents. - [DYNAMIC_CONTEXT_INJECTION]: The
scripts/create-adr.mdfile utilizes dynamic context injection (!syntax) to auto-populate the date, git author, and calculate the next sequential ADR number. These operations are read-only project metadata lookups and do not incorporate user-supplied input into the shell commands, preventing command injection. - [DATA_EXPOSURE]: The skill reads from the project's documentation directory (
docs/adr/) and git configuration to maintain the ADR index. It does not attempt to access sensitive system files, environment variables, or credentials. - [INDIRECT_PROMPT_INJECTION]: As a tool designed to ingest user-provided architectural context and generate files, the skill naturally has a data ingestion surface. However, it implements proper sanitization for file paths (using regex slugs) and is intended for local documentation workflows, posing a negligible risk in its intended context.
Audit Metadata