feat(autoreview): support cursor-agent engine

[hxy91819]

What Problem This Solves

The docs repo's hardened autoreview helper did not support Cursor's headless cursor-agent engine, so the docs workflow could not use the same Cursor-based review path added to the main repo.

Why This Change Was Made

This adds cursor-agent as an explicit opt-in autoreview engine using the docs helper's hardened engine environment. The integration keeps the reviewed repo out of Cursor's trusted workspace, passes the prompt over stdin, rejects unsupported Cursor flags, preserves streaming display, maps hyphenated engine names to environment defaults such as AUTOREVIEW_CURSOR_AGENT_MODEL, and adds JSON extraction for Cursor responses that include explanatory text before the structured report.

User Impact

Docs contributors can run autoreview with --engine cursor-agent or AUTOREVIEW_ENGINE=cursor-agent. Existing engines and the default review behavior are unchanged.

Evidence

• python -m py_compile .agents/skills/autoreview/scripts/autoreview .agents/skills/autoreview/scripts/test-review-harness.py
• python .agents/skills/autoreview/tests/test_autoreview_hardening.py
• .agents/skills/autoreview/scripts/autoreview --self-test-engine-isolation
• .agents/skills/autoreview/scripts/autoreview --self-test-config-defaults
• .agents/skills/autoreview/scripts/test-review-harness --fixture benign --engine cursor-agent
• .agents/skills/autoreview/scripts/test-review-harness --fixture malicious --engine cursor-agent
• AUTOREVIEW_ENGINE=cursor-agent .agents/skills/autoreview/scripts/autoreview --mode commit --commit HEAD
• git diff --check

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.