muxy-extensions
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill employs the
muxyCLI tool to perform various development operations, including building, loading, and testing local extensions. These commands are essential for the primary purpose of the skill and are focused on local workspace management. - [EXTERNAL_DOWNLOADS]: The skill references standard package management via
npm installand recommends the installation of the GitHub CLI (gh) from official sources. These are well-known developer tools and services. - [INDIRECT_PROMPT_INJECTION]: The skill has the capability to read and capture the state of local extension web pages using commands like
snapshotandread. While this creates an attack surface for indirect prompt injection from content within the user's own extension, the risk is inherent to the development and debugging workflow. - Ingestion points: Content is ingested from local pages via
muxy extension page snapshot,muxy extension page read, andmuxy extension page eval(SKILL.md). - Boundary markers: No explicit delimiters or "ignore" warnings are provided for the content read from pages.
- Capability inventory: Shell command execution via
muxyandgh, and local file system access for extension development. - Sanitization: Not specified for the content extracted from the debugged pages.
- [DYNAMIC_EXECUTION]: The
muxy extension page evalcommand allows the agent to execute JavaScript within the context of a development extension. This is a standard debugging feature used to inspect the state of the user's local code.
Audit Metadata