transfer-context

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Python subprocess module to execute local git commands for repository state collection. These executions are constrained to specific arguments and utilize absolute paths or system-resolved paths, preventing arbitrary command injection.\n- [DATA_EXPOSURE]: The implementation enforces strict filesystem permissions, setting checkpoint directories to mode 0700 and files to 0600. This ensures that session state data remains private to the executing user.\n- [CREDENTIALS_UNSAFE]: An integrated secret scanning engine uses regex patterns to detect and block common sensitive markers (such as API tokens and private keys) before they are written to a checkpoint, mitigating the risk of accidental credential leakage.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from repository files and transcripts. It mitigates potential injection risks by providing clear instructions to the agent to treat this content as data rather than instructions, and by using boundary markers to separate system-generated state from model content.\n- [REMOTE_CODE_EXECUTION]: The skill operates entirely within the local environment. It contains no mechanisms for fetching remote scripts or performing network requests, adhering to a local-only security model.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 11:36 PM