sol-loop
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes
gitandgh(GitHub CLI) commands to inspect the local repository state, fetch remote changes, and retrieve pull request metadata. These operations are standard for a developer-oriented tool and are necessary for the skill's stated purpose of automating code reviews. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from local code files and GitHub Pull Requests, which are then transmitted to a subagent ("Sol") for review. This creates a surface for indirect prompt injection if the source data contains malicious instructions.
- Ingestion points: The skill ingests data using
git diff,git status, andgh pr viewas described in Step 1 and Step 2 of the workflow. - Boundary markers: The skill provides a structured prompt brief (Step 2) to the reviewer subagent to define its scope and constraints, though it does not employ explicit data delimiters like XML tags for the code content itself.
- Capability inventory: The skill uses
delegate_taskto interact with subagents and has the capability to modify local files to apply suggested fixes. - Sanitization: No specific filtering or sanitization of the ingested code or PR metadata is performed prior to sending it to the reviewer subagent.
Audit Metadata