codex-chatgpt-control-bridge

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow reads OUTSIDER-authored free text from the visible ChatGPT web session (e.g., via askInThread/runner.run capturing Markdown responses from chatgpt.com), which is not authored by the operating user and is ingested into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The Python example runs npx --yes --package codex-chatgpt-control codex-chatgpt-control-backend which will fetch and execute remote code from the npm package (see https://www.npmjs.com/package/codex-chatgpt-control), i.e., a runtime dependency that executes remote code.