01
Agent execution loop
Build context, call models, execute tools and write results back until work is complete.
Back home
JOYHOUSEBOT AGENT
An extensible agent foundation built to run and remember
A Python-native, OpenClaw-compatible open-source foundation connecting models, channels, tools, knowledge and automation through one architecture.
View open-source project$ joyhousebot agent
loading context · memory · tools
connecting channels · knowledge · automation
JOY › Ready to understand the task, use tools and bring the result back.
02
Multi-channel, multi-agent
Connect Telegram, Discord, Slack, Feishu, DingTalk, email and more.
03
Memory and knowledge
Layered memory, full-text search and vector retrieval maintain long-term context.
04
Safer execution
Schema validation, permissions, secret filtering and optional Docker sandboxes.
Let every interface connect to the same you
Extension, Agent and future hardware share one direction: understand information, take action and bring long-term data home to JoyHouse.