cmux-customization

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses standard Unix utilities such as sed, cp, test, and date to inspect, back up, and manage configuration files in the user's ~/.config/cmux/ and project-local .cmux/ directories.
  • [DYNAMIC_EXECUTION]: The skill configures cmux actions and workspace definitions that execute shell commands (e.g., bun dev, lazygit, ssh, gh). This is the primary intended functionality of the skill for automating developer workflows within the cmux application.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes data from external configuration files which could contain untrusted instructions.
  • Ingestion points: Reads cmux.json and dock.json from ~/.config/cmux/ and .cmux/ file paths.
  • Boundary markers: No specific boundary delimiters are used for the file content.
  • Capability inventory: The skill allows the definition of shell commands that are subsequently executed by the cmux runtime.
  • Sanitization: Risks are mitigated by the requirement to use the cmux-settings validate tool and to perform JSON/JSONC parsing checks before reporting completion.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 02:17 AM