cmux-workspace-builder

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly asks for environment variables and commands and then instructs the agent to generate a complete cmux.json including env key/value pairs and command strings, which would require the LLM to include secret values verbatim in its output.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill's Step 0 explicitly requires fetching the live documentation at https://cmux.com/docs/custom-commands before doing anything else, so the agent will ingest and act on open/public third-party content (the cmux docs) that can materially change how it generates configs.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly instructs the agent to fetch https://cmux.com/docs/custom-commands at runtime and use that documentation to determine field names and behaviors for generating the cmux.json, so external content would directly control the agent's generated prompts/schema and is a required dependency.