working-with-jj

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: Several helper scripts (jj-batch-desc, jj-checkpoint, jj-desc-transform, jj-show-desc) execute shell commands and the jj binary to perform version control operations and metadata retrieval.
  • [DYNAMIC_EXECUTION]: The jj-desc-transform script executes a user-provided command passed as arguments to process revision descriptions, which allows for dynamic execution of shell commands based on inputs.
  • [INDIRECT_PROMPT_INJECTION]:
  • Ingestion points: Revision descriptions are read from the repository using jj log -T description in multiple scripts (jj-batch-desc, jj-desc-transform, jj-show-desc) and reference documentation (references/batch-operations.md).
  • Boundary markers: The instructions do not define boundary markers or safety warnings for handling potentially untrusted repository content.
  • Capability inventory: The skill includes scripts that can execute arbitrary shell commands (jj-desc-transform) and modify repository state (jj-batch-desc, jj-desc-transform).
  • Sanitization: There is no sanitization of the revision descriptions before they are piped into transformation commands or written back to the repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:43 PM