A hosted remote-MCP service that lets AI agents operate Feishu/Lark, reusing lark-cli's 200+ tools and 20+ Skills. One Feishu app shared across the team, centrally managed by IT; members authorize once in the browser, data isolated per user. Built on AWS Bedrock AgentCore.
aws-samples/sample-lark-mcp-on-agentcore is a hosted remote MCP service that lets AI agents use Feishu/Lark through lark-cli's tools and multi-step workflow skills. Teams can deploy one centrally managed service while individual members authorize their own Feishu identities, with the service running on AWS Bedrock AgentCore. The catalogue add-ons help agents perform Lark operations through this service.
Latest release v1.17.0+larkcli.1.0.97 — v1.17.0 — lark-cli 1.0.97: release approval, idle-reclaim fix · 10 Oct 2026
2 files for Codex, OpenCode and Claude Code: sample-lark-mcp-on-agentcore AGENTS.md, sample-lark-mcp-on-agentcore CLAUDE.md — 1,848 tokens loaded in every session.
AGENTS.md A 1,843 tok CLAUDE.md A 5 tok These files are aws-samples/sample-lark-mcp-on-agentcore's own configuration — they tell Codex, OpenCode and Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.