Changelog

Have something to say?

Tell us how we could make the product more useful to you.

Pinned

Connect Capy AI to your WeChat

Connect Capy AI to your WeChat and turn WeChat into an AI chat window! Features: • Chat with Capy AI directly in WeChat (powered by claude-sonnet-4.6) • Independent accounts for each user, with no interference between them • Supports both text and voice-to-text messages in WeChat • Chat history is stored only in memory and is cleared upon closing, ensuring security and reliability Installation Guide (3 Steps): Step 1: Have Capy install the skill In the Capy chat window, type: /install capytowechat Or simply say: "Connect WeChat and Capy" "WeChat bot" "Integrate WeChat with AI" Step 2: Scan the QR code to link your WeChat Capy will display a QR code image. Scan it with the WeChat app on iOS → Tap "Confirm." (The QR code automatically refreshes every 35 seconds, so don’t worry about it expiring.) Step 3: Start Chatting After successful binding, Capy automatically starts the service. Open WeChat and send a message directly to the account you just linked—the AI will reply instantly! Security Guarantees: • Your WeChat token is stored only locally in ~/.capy/wechat/account.json (permissions 0600) • The AI key is read only from environment variables and is not written to any files • Message content is not logged; only metadata (user ID prefix, message length) is recorded Technical Architecture: WeChat → Official ClawBot ilink API → capy-wechat long-polling service → Capy AI Gateway → Reply Skill File download (copy to your Capy workspace and install): capytowechat.skill (approx. 10KB, includes full source code) Requirements: WeChat for iOS + Capy Sandbox (AI_GATEWAY_API_KEY is automatically configured)

tom tom 25 days ago

1
🌟

Show & Tell

Pinned

In Progress

🦫 Share Feedback, Report Bugs, and Show What You Built 🦫

We just launched our Happycapy Feedback, a better way for you to share ideas, report issues, and show what you’re building with Happycapy. To help us respond faster and keep things organized, please choose the right tag when posting: - 🐛 Bug Report → Something not working as expected? Let us know what broke. - 💡 Feature Request → Got an idea or something you wish HappyCapy could do? Tell us. - ❓ Question → Not sure how something works? Ask anything about usage here. - 🧩 Skill Feedback → Want a new skill, integration, or found issues with an existing one? Drop it here. - 🌟Show & Tell → Built something cool? We’d love to see it. If you’re looking for guides or how-to instructions, check out our Docs: https://docs.happycapy.ai/ Thanks for helping us make Happycapy better. We’re building this with you ♥️

min zhou 27 days ago

🌟

Show & Tell

ᅟ✧ᅟ⦻ᅟ✢ᅟꖅᅟ✸ᅟ✧ᅟ𖧷ᅟ✧ᅟ✸ᅟꖅᅟ✢ᅟ⦻ᅟ✧ᅟ◦୦◦◯◦୦◦⠀       ⠀◦୦◦◯◦୦◦ᅟ✧ᅟ⦻ᅟ✢ᅟꖅᅟ✸ᅟ✧ᅟ𖧷ᅟ✧ᅟ✸ᅟꖅᅟ✢ᅟ⦻ᅟ✧ᅟ

⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀   ⰙИNITIᖆᖇᗯᖆᖇITIИNⰙ   ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀    ⰙNИIꞱIᖈᖉᗰᖈᖉIꞱINИⰙ    ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ​⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀         ᖆᖇᑌOYOᑌᖆᖇ         ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀          ᖈᖉᑎO⅄Oᑎᖈᖉ          ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ​⠀⠀⠀⠀⠀       ƎE𑪼𑪽YᖆᖇTƎEMYƧSYMƎETᖆᖇY𑪼𑪽ƎE       ⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀        ƎE𑪽𑪼⅄ᖈᖉꞱƎEꟽ⅄SƧ⅄ꟽƎEꞱᖈᖉ⅄𑪽𑪼ƎE        ⠀⠀⠀⠀⠀ ​⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀         ᖆᖇᑌOYOᑌᖆᖇ         ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀          ᖈᖉᑎO⅄Oᑎᖈᖉ          ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ​⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀   ⰙИNITIᖆᖇᗯᖆᖇITIИNⰙ   ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀ ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀    ⰙNИIꞱIᖈᖉᗰᖈᖉIꞱINИⰙ    ⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀

𖣠⚪𔗢⚪🞋⚪𔗢⚪𖣠◦୦◦◯◦୦◦⠀⠀⠀⠀⠀⠀◦୦◦◯◦୦◦𖣠⚪𔗢⚪🞋⚪𔗢⚪𖣠 10 minutes ago

💡

Feature Request

How can I estimate my credit in the

Attched my billing info above. Yes, upgraded to Max just AFTER max+ disappeared. [crying…] Now my question is, I can only see my left credits in the subscription page, but have totally no idea when will it get refreshed and recharged. Is it a nature month? or from the day I updated to Max plan? This is important to me cause I only get half credits left 3 days after upgrading to max. 😂

Joey Yin about 7 hours ago

❓

Question

Concern Regarding New Weekly Usage Limits on Max Plan

Hi Happycapy Team, I’m writing to express my strong dissatisfaction with the recent change to the Max Plan usage policy. When I originally purchased the plan for $200, it was clearly communicated as “unlimited.” However, the introduction of a weekly credit limit (30,000 credits) significantly changes the nature of what was promised. This feels misleading and undermines trust, especially for users like me who rely heavily on your product for ongoing workflows. The current weekly limit is particularly problematic. Many real-world workflows operate on a daily basis, not weekly cycles. With the new system, I’ve already hit the limit by Thursday, which effectively blocks my ability to continue working for the rest of the week. This creates serious disruption and makes the product far less usable. I’d like to propose a more practical alternative: Instead of a weekly limit, consider implementing a daily usage limit that resets every 24 hours, similar to how Claude handles usage. This approach would: Better align with real user workflows Prevent complete mid-week lockouts Provide more predictable and consistent access Right now, the weekly cap creates a poor experience where usage becomes front-loaded and then completely unavailable, which is not sustainable for active users. I hope you take this feedback seriously and reconsider the structure. I genuinely like the product, but this change significantly impacts its usability. Looking forward to your response.

iuweudhoue 6 days ago

2
🐛

Bug Reports

API Error 400: Invalid signature in thinking block during session interactions

Dear Support Team, I am experiencing a recurring API error during my session interactions with the platform, and I would appreciate your assistance in resolving this issue. Error Message: API Error: 400 messages.21.content.0: Invalid signature in thinking block Issue Description: This error occurs consistently under the following circumstances: When the conversation context becomes longer - After multiple exchanges in a session, the error starts appearing After switching models - Specifically when switching from MiniMax to Sonnet model during an active session Impact: The error prevents me from continuing the conversation and disrupts my workflow, requiring me to restart the session. Expected Behavior: The system should handle longer contexts and model switches gracefully without throwing validation errors related to thinking block signatures. Additional Context: The error references "messages.21.content.0", suggesting it may be related to how thinking blocks are being formatted or validated when context grows or models change The issue appears to be related to thinking block signature validation in the API request processing Request: Could you please investigate this issue and provide guidance on: Whether this is a known limitation or bug Any workarounds available while a fix is implemented Expected timeline for resolution if this is a known issue Thank you for your attention to this matter. Please let me know if you need any additional information or logs to help diagnose the problem. B.R. Jesse Zhu

Jesse Zhu 16 days ago

1
❓

Question

Feedback on Weekly Limits

I checked out the new weekly limit feature: Here’s some feedback—I just had to get this off my chest: 1. Even if there’s a limit, make it a bit higher. Last week, after using it normally for three and a half days, I checked on Saturday and saw that the new weekly limit had already hit 100%. It’s simply not enough—it doesn’t even reach the maximum. 2. Looking at the weekly limit settings, it seems to be limiting the number of interactions with Capy. This feels unreasonable. If the goal is to prevent cheating, limiting by IP, MAC address, or model usage makes sense—but limiting by interaction count doesn’t. From a cost perspective, using M2.7 to handle basic operations is sufficient, but who can guarantee the interaction count? We’re already max users; at the very least, I don’t want to have to worry about prompt limits. It really ruins the experience.

Jason 25 days ago

2
🐛

Bug Reports

Feature Request: Automations 支持"智能新开 Session"模式

1、问题描述 当前 Automation 运行时会复用指定 Desktop 中的已有 Session。如果该 Session 之前有过大量交互（数百甚至上千轮对话），Automation 触发的任务会继承这些历史上下文，导致： a、执行中途触发 Context Compaction：长对话积累到上下文窗口上限时，系统会在任务关键节点强制压缩上下文，导致任务中断、状态丢失、执行时间显著增加 无法从 Agent 内部自救： b、Agent 运行在 Session 内部，没有能力新建 Session 或切换 Session，只能被动承受上下文污染的后果 2、期望功能 在 Automation 配置中，Session 策略除了当前的"复用当前 Session"外，增加一个选项： 【智能新开 Session 模式】 运行前先检查当前 Session 的上下文环境，当检测到新任务可能会触发 Context Compaction 时，自动新建一个全新的 Session，然后在新 Session 里执行 Automation 任务。如果上下文空间充足，则继续在当前 Session 中执行。 补充：这个功能对所有多步骤、长时间运行的 Automation 任务都有价值，能增加automation成功的概率，并减少执行时间

Jason 25 days ago

1
💡

Feature Request