do-pr-reviewlisted

# PR Review Review a pull request by analyzing its changes against the plan, checking code quality, validating tests, and capturing screenshots of UI changes. ## Surface Screenshot capture in this skill runs against the user's real, logged-in Chrome via BYOB MCP (`mcp__byob__browser_*`). Public preview deploys and authenticated staging URLs are screenshotted the same way — BYOB just shows you the page the user would see. There is no anonymous-headless fallback; that surface was retired in #1256. The calling session must have `requires_real_chrome=True` set. For SDLC pipeline runs, the bridge auto-infers from message content; for manual review runs, pass `valor-session create --needs-real-chrome ...`. Two concurrent real-Chrome sessions race on the active tab. ## Review Identity ### Bot Account Model (opt-in per machine) Pipeline-driven reviews MAY post under a dedicated service-account identity instead of the operator's `gh` credential. The bot identity is **opt-in per machine**: only the dedicated bot machine sets `SDLC_AGENT_GH_TOKEN`. Standard machines leave it blank, and pipeline reviews post under the operator's `gh` credential — reviewing our own commits is the accepted default. **Environment variables:** | Variable | Purpose | When set | |----------|---------|----------| | `CLAUDE_AGENT_REVIEW` | `1` when running in pipeline context | Set by `sdk_client.py` at session spawn | | `SDLC_AGENT_GH_TOKEN` | PAT for the bot account (e.g. `yudame-sdlc-bot`) | Optional