integrate-opencut
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the cloning of the OpenCut repository from GitHub and the installation of development tools (Moon, Proto, Bun, Rust) from their respective official sources.
- [COMMAND_EXECUTION]: Defines a mapping of developer intents to specific shell commands using the
moontask runner for building, testing, and developing the project locally. - [INDIRECT_PROMPT_INJECTION]: As the skill interacts with external repository content, it possesses an attack surface for indirect prompt injection if the source files were compromised.
- Ingestion points: Source code is ingested into the local environment via
git cloneas specified inreferences/integration.md. - Boundary markers: The skill does not use technical delimiters for repository content but provides strict guidelines on what the agent should and should not do with the data.
- Capability inventory: The skill allows execution of build tasks (
moon run) and toolchain management (proto use), which could execute logic found within the repository. - Sanitization: This risk is mitigated by pinning the repository to a specific, reviewed Git commit hash (
4d8c49ed0706c4dc145361e01c6b1f1a87cbb863), preventing the agent from automatically pulling unvetted changes from the upstreammainbranch.
Audit Metadata