atomic-code-edits
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a Python implementation (
apply_atomic_patch) that performs file system operations, specifically reading from and writing to files. This is the intended and transparent behavior of the skill for performing localized code replacements. - [PROMPT_INJECTION]: The skill contains an "Agent Operational Directive" that instructs the agent to use specific tools for editing tasks. This is a legitimate instructional constraint aimed at optimizing performance and reliability, rather than an attempt to bypass security guardrails.
Audit Metadata