setup-context-driven
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is centered around the execution of the
roundfixCLI tool (e.g.,roundfix baseline update,roundfix baseline apply). This tool has the capability to inspect the repository and perform file system mutations based on Baseline Profiles. - [REMOTE_CODE_EXECUTION]: The skill provides instructions for restoring external skills via the
roundfix baseline skills restorecommand. This mechanism allows the agent to download and integrate external logic into the repository's 'Skill Set'. While the skill mandates a human-in-the-loop confirmation via a Plan Digest, it establishes a workflow for remote code acquisition. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of 'Baseline Profiles' and 'Repository-Specific Normative Rules'. If these external or repository-resident files are compromised, they could contain instructions that influence the agent's behavior during the setup and planning phases. The skill lacks explicit prompt-level sanitization or delimiters for these inputs, though it mentions logical boundaries enforced by the CLI tool.
Audit Metadata