create-adr
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands such as
lsor glob patterns to scan the local filesystem for ADR directories (e.g.,docs/adr/,adr/) and to determine the next available sequential number for new documents. This is an expected and benign behavior for managing document history. - [INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text for architectural decisions and places it into markdown templates, creating a theoretical attack surface if the generated documents are later parsed by other agents.
- Ingestion points: User input gathered in
references/process.md(Gather Context step). - Boundary markers: Absent; user text is interpolated directly into markdown templates without specific delimiters.
- Capability inventory: File system write operations defined in
references/process.md(Step 5). - Sanitization: Absent; the skill does not perform sanitization of user-provided content for embedded instructions.
Audit Metadata