🧞‍♂️ PicoClaw Wishlist: What features do you want next?

[Zepan]

Welcome to the PicoClaw Ideas section! 👋

Think of this category as our community Wishlist. We are building a lightweight AI agent for low-cost hardware, and we want to know what you need most.

How to use this:

Search first: Check if someone has already posted your idea.

Upvote (⬆️): If you see an idea you like, please vote for it! We prioritize features with the most votes.

Post new ideas: Don't be shy! Whether it's support for a specific board or a new software capability, let us know.

Example requests:
"Add multi-agent"
"Support for swarm mode"
"Integration with Home Assistant"
Let's build the best tiny AI agent together! 🚀

[czyt]

• 支持server酱的bot
• gateway web UI
• 直接调用claude、opencode、codex作为子agent
• 支持自定义的OpenAI提供商，目前提供商都是写死的。实际上只需区分api类型老版本的msg、claude格式api、response格式api

[kaldune-cmd]

I wish PICO CLAW would acknowledge the X community that raised over $5,000 dollars for them and is ready for transfer on github. That would be fantastic

[czyt]

i have made a new project called nekobot https://github.com/ca-x/nekobot

[vangoft]

Add multi-agent mode
add desktop control for those using it on an actual desktop environment so it can run fully autonomously (at your own risk of course)
add support for Codex, Claude Caude, Gemini subscriptions

[Erildo]

agree, let him control more, at our own risk

[imziza]

pls add the gemini cli auth

[ashishk1331]

Whatsapp integration?

[dmatosjunior12-hub]

Sim, eu apoio (e preciso)

[mrbeandev]

For anyone looking for easier deployment and better provider control, I've just updated PR #239 with some major additions:

1. Coolify Support: Added a dedicated Dockerfile.coolify, docker-compose-coolify.yml, and an automated entrypoint script. This makes it a one-click deploy on VPS platforms like Coolify.
2. Custom OpenAI Providers: Added support for PICOCLAW_CONFIG_JSON. You can now define any custom OpenAI-compatible provider (Groq, OpenRouter, local models) with its own BaseURL and APIKey entirely through environment variables.
3. Advanced Telegram UI: Added a dynamic /models list that shows what's actually available across your providers, and a /model provider/model command to switch between them instantly.

Hopefully this covers some of the most requested features here! 🚀

[adamkoncz]

Custom skill, tool and hook support. Better and up to date documentation.

[harshbansal7]

Custom Skill support with integration of Clawhub.ai is in dev.

[spacepirate0001]

#255 OpenTelemetry Observability for PicoClaw (Edge/Agent Use Cases)

[dligthart]

Will you continue to take features from the openclaw project so picoclaw stays in sync?

[dligthart]

Multi-agent development team with predefined roles.

[A2-NieR]

Support for privacy focused https://venice.ai/
https://docs.venice.ai/api-reference/api-spec

[JokerQyou]

• Interrupt-able conversation. Currently picoclaw replies to every single message, which is not ideal. For example I might forward a message from public channel to picoclaw, and ask what it thinks about this message. In this case picoclaw would always reply two separate messages. Ideally the second one should interrupt the agent, and force it to merge operations.
• Map sessions to topics. This is specific to Telegram. Bots can enable topics, and in each topic the same user might be talking about things that are totally different. I'd like the agent to have separated sessions mapped to each topic. This is currently not possible.
• Telegram bot commands for session management, like /new to start a new session.