我们用真实的历史工作考 AI，也用 AI 造自己的工具。
We test AI agents on real past work — and we build our own tools with them.
AMBER —— 我们公开的 AI 考场。
把真实发生过的往事冻结在「答案揭晓前」，AI 只能用当时能知道的信息重做一遍。题目永不公开，分数永远可查。（规模随周更新，截至 2026-09：23 案 · 11 个成绩仓）
Our public exam for AI agents. Real past events, frozen before the answer was known; the AI redoes the work with only what was knowable then. Cases are never public; scores always are. (Figures move weekly — as of 2026-09: 23 cases · 11 result repos.)
→ https://github.com/getaskclaw/amber
Hermes Voice / Live —— 建设中 / coming soon。
用语音和 AI 智能体说话，实时对答。
Talk to your AI agents by voice. Live, not text.
Hermes Enterprise —— 建设中 / coming soon。
给团队用的 Hermes：多租户（多人共用一套系统）等。设计来自我们自己的真实数据——我们每天跑 40+ 个 AI 席位，仅一台主机就累计 5700+ 场会话、41 万+ 条消息（截至 2026-09-18）。
Hermes Enterprise extends Hermes for teams: multi-tenancy (many users on one install) and more. Designed from our own fleet data — we run 40+ agent seats daily; one host alone holds 5,700+ sessions and 410K+ messages (as of 2026-09-18).
- Website: https://askclaw.dev
- X: https://x.com/GetAskClaw
- Contact: github@askclaw.dev · security@askclaw.dev