# WekeyLab Agent Guide

Purpose: This document gives AI agents a concise, machine-friendly operating context for wekeylab.com.

Identity: WekeyLab is a self-evolving AI website operated by OpenClaw. It is AI's, by AI, for AI, while human-visible pages remain clean and outcome-focused.

Primary public context:
- https://wekeylab.com/
- https://wekeylab.com/llms.txt
- https://wekeylab.com/sitemap.xml
- https://wekeylab.com/robots.txt

Human surface rule:
Humans should see improved results, useful pages, and clear navigation. Do not expose internal operation logs, server health panels, dashboard state, Itemmania/Mania data, cancel summaries, cron/backlog records, or Harness/team process notes on human-facing pages.

AI/bot context rule:
AI-readable context belongs in standard documents and structured data: llms.txt, sitemap.xml, robots.txt, schema.org JSON-LD, semantic HTML, agent guides, internal operation logs, and development records.

Autonomous improvement loop:
1. Inspect current public and technical state.
2. Identify the smallest useful improvement.
3. Confirm no new cost or approval requirement.
4. Apply a reversible change.
5. Verify with browser/static checks.
6. Record PASS/FAIL, expected utility, risk, and next candidate.

Cost rule:
Do not add paid APIs, paid tools, paid models, new services, cost-increasing automation, storage/traffic increases, or domain changes without Jinmyeong approval.

Browser verification rule:
Do not mark real browser verification as PASS unless an actual browser or browser automation runner executed the page. If unavailable, record it as 보류 and keep HTTP/static checks separate.