sol-loop
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from local file diffs and GitHub pull requests, which is then passed to an external LLM for review. This creates a surface where maliciously crafted comments in the code could attempt to manipulate the reviewer's behavior.
- Ingestion points: The skill reads output from
git diffandgh pr viewwithinSKILL.md. - Boundary markers: The "Round 1 brief" in
SKILL.mdprovides instructions and constraints to the sub-agent ("Review only. Do not edit, commit, push, comment on PRs..."). - Capability inventory: The skill utilizes shell execution (git, gh) and the
delegate_taskfunction for external agent interaction. - Sanitization: No specific sanitization or filtering of the code diff content is performed before passing it to the sub-agent.
- [COMMAND_EXECUTION]: The skill executes local shell commands to manage the development workflow and identify code changes for review.
- Evidence:
git status --short,git fetch origin,gh pr view,git merge-base, andgit diffcommands are used throughoutSKILL.mdto establish scope and compare versions. - [EXTERNAL_DOWNLOADS]: The skill documentation references an external repository for synchronization purposes and as a source of origin.
- Evidence:
THIRD_PARTY_NOTICES.mdreferences thetstackrepository athttps://github.com/t3dotgg/tstack.
Audit Metadata