timesfm
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
One API key for 6 AI models. Pay-per-use. MCP protocol support with web search.
Skills that teach the AI this kind of work; they work alongside this one
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.
Open-source intelligence for the global theater. Track everything from the corporate/private jets of the wealthy, and spy satellites, to seismic events in one unified interface. Hook an AI agent up to have it parse through data and find previously unseen correlations. The knowledge is available to…
scrape data from Google Maps. Extracts data such as the name, address, phone number, website URL, rating, reviews number, latitude and longitude, reviews,email and more for each place
Lists are ordered by a usage score (out of 100) built from these parts
| Popularity | 6/80 | GitHub stars or weekly downloads: the larger counts 60, the smaller 20 |
|---|---|---|
| Freshness | 7/12 | Full marks for a release or code change within 30 days |
| Still answering | 0/8 | The cloud endpoint answered at the last check |
| Usage score | 13/100 |
Quality: Passed 2/4 basic checks (shown, not used for ordering) · Momentum: No data yet (Used only on the Trending tab)
Pick the AI app you use, then copy the settings.
Passed 2/4 checks · Runs on your computer
We hid some of the commands
Connect by URL: the server address did not answer when we checked
No install path has passed yet. See the maker’s instructions or pick another server in this category · Maker’s install instructions · Others in this category
First: which AI do you use?
Pick your app and the steps appear below.
Claude cannot connect ATEX AI Gateway yet
No install path has passed yet. See the maker’s instructions or pick another server in this category
ChatGPT cannot connect ATEX AI Gateway yet
No install path has passed yet. See the maker’s instructions or pick another server in this category
Claude Code cannot connect ATEX AI Gateway yet
No install path has passed yet. See the maker’s instructions or pick another server in this category
Cursor cannot connect ATEX AI Gateway yet
No install path has passed yet. See the maker’s instructions or pick another server in this category
VS Code cannot connect ATEX AI Gateway yet
No install path has passed yet. See the maker’s instructions or pick another server in this category
No install path has passed yet. See the maker’s instructions or pick another server in this category
Based on how this server connects and on each app’s own docs, checked Nov 10, 2026.
Connect over the internet in
GooseLM StudioMsty StudioCherry StudioVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatOpen WebUIRaycast
The server did not answer when we last checked, so we could not read its tools.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 19, 2026
The registry name was verified with the GitHub account lm203688, which owns the repo
Code last changed 30 days ago
No licence file found in the repo
Did not respond within 6 seconds when we checked
Did not respond within 6 seconds when we checked
Not checked yet; we read the tool list when we check the endpoint
The cloud address answered 0 of the last 5 checks · no answer last time · checked Nov 19, 2026
If you made this, paste this code into your README. The badge links back to this page.
[](https://mcp.popnx.com/en/servers/lm203688-atex-ai-gateway)Ask, share a tip or report that it does not work. Sorted by what people marked helpful.
No posts yet. Be the first to ask or share.
Posts here are written by users, not the PopNx team.
No reviews yet
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 2 of 4 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs in the maker’s cloud. Connect over the internet with nothing to install, but what you send passes through the maker’s servers.
The GitHub repo last changed 30 days ago, and the latest version (4.3.0) reached the registry 6 months ago.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. It connects by URL, so apps that accept remote MCP servers work too, such as ChatGPT on the web (full MCP on Business and above; Pro read-only).
In the same category: OpenSEO, Exa, Microsoft Learn MCP, Docs, COTAL actions (cotal.ai)
Same category, other makers, commands already checked. Official ones first, then by stars.
Reading a card:↓ Downloads per weeki★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
Keyword research, site audits, rank tracking and AI visibility for your AI agent.
Connect AI agents to Exa for web search, content fetching, and multi-step research.
Official Microsoft Learn MCP Server – real-time, trusted docs & code samples for AI and LLMs.
Search and read Empryo's documentation. Read-only, no auth, no local access.
cotal.ai actions: product overview, site search, build log, feedback, Cloud waitlist, updates, calls
| This one ATEX AI Gateway | OpenSEO | Exa | Microsoft Learn MCP | Docs | COTAL actions (cotal.ai) | |
|---|---|---|---|---|---|---|
| Runs where | Runs in the cloud, nothing to install | Runs in the cloud, nothing to install | Runs in the cloud, nothing to install | Runs in the cloud, nothing to install | Runs in the cloud, nothing to install | Runs in the cloud, nothing to install |
| Needs a key | No | No | No | No | No | No |
| Add with + in Claude | No | Yes | Yes | Yes | No | No |
| Basic checks | 2/4 | 5/5 | 2/4 | 4/4 | 4/4 | 4/4 |
| License | – | MIT | MIT | CC-BY-4.0 | NOASSERTION | Apache-2.0 |
| Price | No cost listed in the registry | No cost listed in the registry | No cost listed in the registry | No cost listed in the registry | No cost listed in the registry | No cost listed in the registry |
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
合规版积分预测社区(类 Manifold × Good Judgment Open × Kalshi 信息呈现)。 FastAPI + SQLite,积分只送不卖 / 不可流通 / 不可回兑 / 彻底去加密,合规优先。
见 deploy/:Dockerfile / docker-compose.yml / nginx.conf / .env.example / realcast.service(systemd) / backup.sh(热备) / healthcheck.sh / 部署运维手册.md。
cd deploy && cp .env.example .env # 填 ADMIN_TOKEN / CORS_ORIGINS
docker compose up -d --build
Railway 提供持久化卷,SQLite 可直接用,无需改 Postgres。配置文件已就绪: railway.json + deploy/Dockerfile.railway(构建上下文为仓库根目录,与 deploy/Dockerfile 不同,别混用)。
控制台操作步骤:
/data (不挂卷 = 容器每次重建数据归零;库目录不存在时应用会拒绝启动并打印排查提示)| 变量 | 值 | 说明 |
|---|---|---|
ADMIN_TOKEN | 长随机串(≥32 位) | 运营后台/审核/结算必填 |
CORS_ORIGINS | 你的 Railway 域名 | 不要留 * |
DB_PATH | /data/platform.db | 镜像已默认,与卷挂载点对应 |
APP_TZ | 8(国内)/ 0(UTC) | 决定「今天是哪一天」的边界 |
UVICORN_WORKERS | 1 | 不可 >1,原因见下 |
SEED_ON_BOOT | 0 | 演示环境可设 1 自动播种(幂等) |
REDIS_URL | 可选 | 配了才能安全扩容到多 worker |
https://<你的域名>/api/health,确认 status=ok为什么 `UVICORN_WORKERS` 必须是 1:未配 REDIS_URL 时实时广播走进程内 MemoryBackplane(app/core/backplane.py),一个 worker 就是一个独立广播域。 worker > 1 时,A 进程的市场状态变更推不到连在 B 进程上的客户端,不同用户看到的 概率会不一致。要扩容必须先配 REDIS_URL 切到 RedisBackplane,并建议同时迁 Postgres。
为什么 `overlapSeconds = 0`:Railway 默认在新旧容器间做短暂重叠实现零停机, 但重叠窗口内两个进程会同时写同一个 SQLite 文件,轻则锁竞争、重则损坏数据。 设为 0 = 先停旧再起新,代价是每次部署几秒中断。
cd app && python tests/smoke.py --fresh # 期望 168/168 全绿
cd app && python tests/uat_customer.py # 期望 90/90 全绿
提示:本机若设了
HTTP_PROXY,测试脚本访问127.0.0.1会被代理拦截(表现为 502),
运行时请加
env -u HTTP_PROXY -u HTTPS_PROXY -u http_proxy -u https_proxy清除。
script-src(v0.7.1 起)与 style-src(v0.7.5 起)均已去掉 'unsafe-inline', 改为每请求生成的一次性 nonce。nonce 只能由后端在渲染 HTML 时注入,因此:
_render_index),不能由 nginx 等静态托管;deploy/nginx.conf 文件头说明)。sigma。排行榜分两类:积分榜(sort=points,默认,反映活跃度)与技能榜(sort=skill)。
技能榜不按单点分排序,而按 Glicko-2 的保守下界 skill_rating - 1.96 × skill_rd:
skill_rd(评分偏差)直接量化「我们对这个人有多不确定」。下注越多 RD 越小; 长期不下注 RD 会回升(见 rating.rd_decay)。correct/resolved 在样本极小时噪声极大——新用户下 2 注 全中就是 100% 直冲榜首,样本一多又暴跌。用区间下沿后,低样本用户自然排后面, 真实水平随样本积累才浮上来。命中率另给 Wilson 置信下界(accuracy_conservative)。s_ij = score_j / (score_i + score_j),满足 s_ij + s_ji = 1:一场比较零和, 不会凭空造分,且比「谁押中谁赢 1 分」保留了幅度与校准信息。provisional: true, ?ranked_only=1 可只看正式榜位。为什么不用 Brier 直接评 `prob_at_bet`:prob_at_bet 是下注时的市场价格(群体 共识),不是用户自报的信念。拿 Brier 评它,评的是「群体共识准不准」,会出现 「押中 50% 热门(0.25) 输给 未中 40% 冷门(0.16)」的反向结论。改用「跑赢共识」的 标准化技能分 (y - p) / sqrt(p(1-p)) 后方向才正:逆势押中 +2.0 > 从众押中 +0.5 > 0 > 冷门未中 −0.5 > 热门翻车 −2.0。
EXTERNAL_REF_ENDPOINT 配置后,可拉取外部公开市场的概率作为参考水位,用于 冷启动播种、「群体预测 vs 外部共识」偏差分析、以及校验自有 Oracle 的合理性。
合规隔离(硬约束,改动前必读):
markets.external_ref_prob / external_ref_json,绝不参与结算—— 结算只读 oracle_log 与自有权威源。smoke 测试有代码级断言守护这条边界 (结算/Oracle 路径出现 external_ref 即失败)。enabled=false,绝不伪造参考值。此前 v0.7.8/0.7.9 的数值市场 API 与技能评级/排行榜后端已就绪,v0.7.10 完成 SPA 前端打通, 做到端到端可用:
market_type=numeric,渲染「数值预测 + 可选 σ 不确定性」输入框, 走 /api/markets/{id}/participate_numeric;分类市场维持原选项选择。在 v0.7.10 产品级 SPA 基础上,按《项目完整性扫描与改进提升计划》P4 路线图补齐:
GET /api/users/{id}/calibration,按 0.1/0.3/0.5/0.7/0.9 概率桶统计「下注概率 vs 实际命中率」,给出过度自信/保守标签与 Brier;前端「我的预测」以 SVG 可靠性图呈现。GET /api/markets/{id}/forecast,统一分类(LMSR 声誉加权)与数值(CRPS 加权共识)两类共识,并附 deviation_vs_external 群体 vs 外部参考偏差分析(外部概率仅参考、绝不结算,与 external_ref 合规隔离一致)。tiers.skill_bonus_reputation),让「准」的用户更快解锁 UGC 发起权等特权;GET /api/users/{id}/tier 新增 skill_rating / skill_confidence / skill_certified,前端展示「🎯 技能认证」徽标。app/automation/ai_predictor.py 与 POST /api/automation/ai_predict。默认 AI_PREDICT_ENABLED≠1 时为安全 no-op;开启后 AI 走与普通用户完全相同的 participate 路径、受同样规则约束,全部标注 is_ai=1,前端明示「🤖 AI 预测」。无外部参考信号时不预测,避免注入噪声。POST /api/oracle/resolve-due 改用 resolve_due_with_human_fallback()——多源结果分歧或 ORACLE_AUTO_SETTLE 未开启(默认)时一律升级 dev 人工裁定工单,绝不自动结算,守住人工兜底红线;单源一致且显式开启 ORACLE_AUTO_SETTLE=1 才自动结算并留痕。原源直连结算保留为 POST /api/oracle/resolve-from-sources。app/mcp/server.py 与 POST /mcp,纯标准库实现无外部依赖。只读工具(list_markets/get_market/get_forecast/get_leaderboard)免鉴权;写工具 submit_prediction 必须带 Authorization: Bearer 且服务端核验 token 与 user_id 匹配后才走普通 participate 路径。质量门禁:
python tests/smoke.py --fresh现为 168/168 全绿(含 P4 + P5 全量断言)。
竞品/市场扫描结论:纯 play-money 预测在亚洲监管面最小、变现不能靠交易抽水,应走「广告 + 品牌众测 + VIP订阅(非买分) + 实物/抽奖(公示概率) + 匿名数据B2B + 引擎白标」复合栈。真测已站对赛道,缺的是游戏循环。本轮按 Nir Eyal Hook 模型(触发→行动→可变奖励→投入)落地四循环:
core/daily.py + 端点 /api/daily/checkin、/api/daily/picks、/api/daily/picks/claim、/api/daily/status。每日签到(连续递增、断签重置,损失厌恶)、每日一猜(3 场精选、完成领 15 积分)、预测连胜展示。前端「游戏中心」一屏。POST /api/lucky/spin 每日免费 1 次,固定概率表(conic 转盘 UI)掉积分/抽奖券/皮肤碎片/连胜加成/近失安慰——Hunt 类爽点,概率后台不可调。core/seasons.py + GET /api/season、POST /api/season/claim。当前赛季 10 级免费轨,XP 来自真实活动(预测/评论/签到/每日一猜/转盘),奖励为积分/抽奖券/勋章,全部公示;付费轨禁用(因不可人民币买分)。core/cosmetics.py 写好外观抽卡(概率公示 + 保底 10 抽 + 单日上限 5 + 仅外观零货币价值),但 LOOTBOX_ENABLED 默认 0 关闭、前端不开入口,等律所意见后一键开启。core/social.py + 端点 /api/squads(建)、/api/squads/{id}/join|leave、/api/squads/mine、/api/squads(列表)、/api/squads/leaderboard、/api/pk/challenge、/api/pk/mine、/api/pk/leaderboard。战队为纯荣誉聚合榜(成员预测活跃度+校准+声誉),不发放/不转移积分;好友 PK 双方对同一市场比准,结算时由平台奖励池发放胜者小额积分+「PK之王」勋章,输家不付出、双方积分不互转(非流通、非回兑)。市场结算时自动连带结算该市场上的所有 PK。全程 18+,不涉及人民币/加密/赌博三要素。core/adwatch.py + GET /api/ads/offer、GET /api/ads/status、POST /api/ads/watch。用户自愿观看(前端占位 30s)→ 领平台积分(默认 20/次、每日上限 3)。链路:广告主付费→平台发积分→用户免费得,积分仍只送不卖。core/sponsors.py + GET /api/sponsors、POST /api/markets/{id}/sponsor(admin)、GET /api/sponsors/{id}/report(admin)。品牌赞助市场(徽标展示),品牌方取该市场匿名群体情绪报告(无 PII);奖品由品牌提供,参照虎扑/腾讯体育「品牌主题联赛」。种子已演示:双十一市场由「星寰电商」赞助。core/vip.py + GET /api/vip/tiers、POST /api/vip/subscribe。三档(免费/支持者 ¥9/Pro ¥29),权益为去广告/专属分析/抢先体验,绝不发放下注积分、绝不 pay-to-win。真实支付需接 xunhupay+ICP,VIP_PAYMENTS_ENABLED 默认 0 关闭,仅暴露档位说明。core/events.py + GET /api/events、GET /api/events/{id}、POST /api/events/{id}/participate、POST /api/admin/events(admin)、POST /api/admin/events/{id}/close(admin)。活动由时间推导状态(进行中/即将开始/已结束),参与领平台积分(每用户每活动限一次,积分只送不卖),关联真实市场作内容聚合展示。种子已演示三档真实主题季:诺贝尔季(🔬)、NBA 揭幕周(🏀)、双十一众测季(🛒),相对 now 生成,部署后始终有进行中/即将开始的活动。前端「游戏中心」新增「📅 限时活动」卡片。质量门禁:
python tests/smoke.py --fresh现为 168/168 全绿(含 P4 + P5 全量断言)。