Autonomous advisor-driven improvement loop. Runs games, reviews replays with Claude strategic analysis against Protoss guiding principles, prioritizes improvements, and executes them via /improve-bot — repeating for hours to iteratively strengthen the bot. Designed for overnight unattended runs.
Autonomous generation-phase improvement loop. Generates a pool of Claude-proposed improvements, fitness-tests each vs the current parent, stacks the winners onto a new snapshot (with an import-check gate), and regression-checks against the prior parent — repeating for hours until the pool or wall-clock budget is exhausted. Designed for overnight unattended runs.
Run a long autonomous session to improve the Alpha4Gate bot's performance along a measurable axis. Supports demo (no mutations), training-only soak runs (learning self-improvement via the daemon), plan-based dev work, and (with --self-improve-code) fully autonomous reactive hybrid loops. Designed for overnight / unattended runs.
Screenshot every Alpha4Gate dashboard tab using Playwright and review for issues. Detects system state (fresh start, SC2 running, advised run active) and adapts review criteria accordingly. Invoke as "/a4g-dashboard-check".
Alpha4Gate wrapper for ui-review-loop. Starts bot + API + frontend, drives the command panel via Playwright, runs three reviewers. Invoke as "/a4g-ui-test --problem '...'".
Triage the findings from the most recent /improve-bot run log into a pickable list, classify each finding by action-type, and emit a ready-to-run /improve-bot invocation for the picked item. Use between /improve-bot runs to decide what to target next without re-reading the full report.