skills/phrazzld/claude-config/commit/Gen Agent Trust Hub

commit

Pass

Audited by Gen Agent Trust Hub on Feb 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes multiple git commands (status, diff, commit, fetch, push) to automate repository management.
  • [COMMAND_EXECUTION]: Invokes pnpm scripts such as lint, test, and build, which are defined in the local project's package configuration.
  • [COMMAND_EXECUTION]: Grants the agent the authority to delete or consolidate files identified as 'cruft' during workspace tidying.
  • [REMOTE_CODE_EXECUTION]: Running pnpm test or build commands can trigger arbitrary code execution defined within the repository's test files or build scripts.
  • [DATA_EXFILTRATION]: Performs git push operations, which transmit local codebase changes and potentially sensitive committed data to a remote origin.
  • [PROMPT_INJECTION]: Subject to indirect prompt injection because it analyzes untrusted git diff outputs and file contents during the categorization and grouping phases.
  • Ingestion points: git status and git diff outputs processed in the 'Analyze' step.
  • Boundary markers: No delimiters or instructions to ignore embedded commands are present when analyzing diff data.
  • Capability inventory: Shell execution (git, pnpm), file system modification (delete), and network transmission (push).
  • Sanitization: No evidence of sanitization for the data processed during analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 27, 2026, 02:26 PM