skills/harshanandak/forge/rollback/Gen Agent Trust Hub

rollback

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to perform git reverts and checkouts via the Bash tool. The provided documentation enforces strict validation patterns for commit hashes and file paths to ensure operations remain within the project scope and do not execute unintended shell commands.
  • [INDIRECT_PROMPT_INJECTION]: There is a potential ingestion point where the skill parses commit messages to update issue statuses. This risk is addressed in the safety documentation, which specifies the rejection of shell metacharacters and strict pattern matching for extracted identifiers to prevent command injection from untrusted git history.
  • [EXTERNAL_DOWNLOADS]: The skill uses bunx to run the forge package from the npm registry. This facilitates the execution of the primary toolset required for the rollback workflow and is a standard method for utilizing Node.js-based development tools from a well-known package registry.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 10:07 AM
Security Audit — agent-trust-hub — rollback