page-load-loop — the sub-50ms optimizer

Optimizes for speed until every page loads under 50ms. Each beat re-runs the same repeatable benchmark, drafts an optimization for the slowest route, and confirms no regression before reporting it. Requires a human setup step (routes, warm-up, baseline) before the first beat can run.

Category: N · Performance · Heartbeat: schedule (weekly, per catalog; source's own cadence is "after each significant change") · Level: L1 (report-only, drafts-not-applies) · Cost: Medium

Source: Loop #12, "The sub-50 ms page-load loop," from Forward Future's Loop Library, original prompt by Matthew Berman.

Quickstart

1. Read LOOP.md — the six parts, the prompt, the limits, the three stops,
   and the source's verbatim prompt.
2. REQUIRED FIRST: fill in page-load-loop-state.md's "Fixed test conditions"
   section (routes, warm-up behavior, baseline capture) — the source names
   this as mandatory setup before any beat runs. Commit it before beat 1.
3. Claude Code: point a weekly Routine at .claude/skills/loop-task/SKILL.md.
   OpenCode: rename opencode.json.example → opencode.json, add the
   AGENTS.md line, wire skills/loop-task.md into a cron + wrapper script.
4. Rehearse once, by hand, before the schedule ever fires:
   "Run the page-load-loop skill now, exactly as the schedule would."
5. Grade the rehearsal with the loop-verifier agent — it re-runs the
   benchmark and checks for regressions on every other route, not just the
   targeted one. PASS → arm the schedule. FAIL → fix the skill, never the
   report.

What's in this kit

Loop Ready checklist

• Provable success condition — the source's own spec, verbatim: "every page loads in under 50 ms"
• Run limit — 20 runs (added by this library)
• Spine written first, committed empty — plus the source's own required setup fields
• Report-only (L1) start — drafts optimizations in a throwaway worktree, applies nothing
• Human gate placed — the source's own pre-loop setup step, plus review of every drafted change
• One log line per beat, no silent runs
• Kill switch — loop-pause-all in loop-budget.md

How this relates to the course

This kit uses Forward Future's Loop #12 nearly unmodified — its stopping condition ("every page loads in under 50 ms") and checker ("the same benchmark and confirm there are no regressions") are used verbatim. The source's own required human gate — defining the metric, routes, test environment, warm-up behavior, and baseline before any agent cycle runs — is exactly this course's "spine written first, committed" discipline (the seven-item minimum), applied to a real external design rather than restated abstractly.

---

Full attribution: resources/sources.md.