ClaudeIM
Run Claude Code via Feishu chat
What Is ClaudeIM?
ClaudeIM is an enterprise communication and workflow integration skill on EasyClaw designed to run Anthropic's advanced coding assistant, Claude Code, directly inside Feishu (Lark) chat interfaces. Instead of requiring developers and operations teams to interact with terminal CLI environments or separate web tools, this skill builds a secure, real-time message bridge — allowing you to trigger code modifications, debug errors, and run automated deployment tasks directly within your active Feishu team chats.
The skill is built for technical teams, software engineers, DevOps managers, and product owners who want to bring collaborative, AI-assisted development and system troubleshooting into their primary company messaging app.
The expected outcome is real-time code and system orchestration via chat: streaming code updates, interactive error debugging within project directories, and rapid execution of automated system tasks securely mediated via a custom Feishu bot.
How ClaudeIM Works
1. Feishu bot webhook registration. The skill registers custom webhooks with your enterprise Feishu Open Platform application, mapping chat messages directly to Claude Code execution scripts.
2. Natural-language command parsing. When a user sends a message (e.g., "@Claude Code: change the database configuration"), the skill parses the command to identify the target action, file paths, and required system configurations.
3. Claude Code execution container. The skill boots a secure, localized terminal execution container with Claude Code CLI active. It feeds the parsed chat context and target repository directories into Claude's environment.
4. Real-time response streaming. As Claude processes the request and writes or modifies code, the terminal stream is captured and piped directly back to the active Feishu chat, displaying real-time progress and code modifications to the team.
5. Interactive error resolution. For debugging requests, the skill analyzes error outputs or code blocks provided in chat, runs diagnostic tests locally, and provides the resolved code with precise inline comments.
Key Features
- Feishu (Lark) integration: Connects Claude Code directly to your company's active team chat channels.
- Real-time code streaming: Streams code outputs and execution logs back to Feishu in real-time.
- Headless CLI execution: Runs Claude Code CLI actions inside secure background containers.
- Interactive debugging: Analyzes and resolves system and code errors directly from chat inputs.
- Workspace-scoped execution: Limits code modification actions strictly to defined local directory boundaries.
- Guided parameter resolution: Proactively prompts users in chat for missing deployment details (such as host, port, or DB name) before running write commands.
What Problems Does ClaudeIM Solve?
1. Collaborative bug fixing in team chat
A developer posts a backend error trace in their team's Feishu dev channel and tags the bot: "@Claude Code: why is this database connection throwing a timeout?" The skill processes the error, scans the local configuration files, identifies that the DB port is incorrectly mapped, and replies in chat with the exact code correction — allowing the whole team to see the diagnosis and solution simultaneously.
2. Updating configurations headlessly via mobile
An operations manager is away from their laptop and needs to update an API host URL in the project config file. They open Feishu on their phone and message the bot: "Change the API host endpoint to api.prod.com." The skill boots the execution container, locates the target config file, modifies the value, and posts a Git-diff screenshot confirming the update.
3. Generating boilerplate code components in channel discussions
During a product feature discussion, a designer asks: "How should we layout the user profile screen?" A developer tags the bot to generate a clean, modern frontend component. The skill writes the code block and streams it directly to the channel, giving both the developer and developer a clear starting point for their discussion.
4. Running deployment tasks safely
A DevOps engineer wants to run a staging database migration. They message the bot: "Run staging db migration." The skill notes that database actions require specific credentials. It prompts for the target database host, port, and security token, validates the inputs, executes the migration script safely, and streams the success logs to Feishu.
5. Onboarding team members with interactive code review
Junior developers can ask the bot to review their draft code blocks directly in Feishu: "@Claude Code: is there a better way to structure this nested loop?" The skill analyzes the draft for efficiency and security, providing optimized, well-commented alternatives in the chat stream, acting as a collaborative peer-reviewer.
Example Workflow
A developer needs to quickly update a server's environment configuration while discussing deployment options with their team in Feishu.
1. They open the Feishu group chat and type: *"@Claude Code: Update port to 8080 in config.ts."*
2. The skill captures the message, runs the local execution script, and scopes the action to the project directory.
3. Claude Code identifies the `config.ts` file, modifies the port variable, and generates a Git-diff output.
4. The skill streams the live terminal logs and the final Git-diff directly into the Feishu chat window.
5. The developer reviews the diff in chat and replies: *"Confirm change."* The file write is committed, and the team is notified.
Configuration updated and reviewed collaboratively in under 45 seconds.
Getting Started with ClaudeIM
Code：Send a request in Feishu -> Claude executes and streams code in real-time
Run Task：Change the database config -> Missing info, provide host/port/db name
Debug：Why this error? -> Claude analyzes and fixes in the project directory
Frequently Asked Questions
What is Claude Code and how does this skill integrate with it?
Claude Code is Anthropic's official command-line interface (CLI) tool that lets Claude interact with your local directory, edit files, and run terminal commands. This skill builds a secure, real-time message handler that pipes Feishu chat inputs directly to Claude Code and streams the terminal outputs back to the chat.
Is my code uploaded to any third-party servers?
The skill executes Claude Code within a secure local container on your system. Text queries and code contexts are sent directly to Anthropic's secure APIs over HTTPS to generate solutions. Your codebase remains private and is never stored on external intermediate servers.
Can I run this bot in private Feishu group chats?
Yes. You can add the custom Feishu bot to any group chat or interact with it via direct message. To trigger actions in group chats, simply tag the bot (e.g., "@Claude Code") at the start of your message.
What happens if Claude Code attempts to run a destructive command?
The execution container applies strict user permissions, preventing the execution of root-level destructive commands. Additionally, any file write or script execution action must be explicitly confirmed by an authorized developer in the chat before being applied.
How do I configure the Feishu bot credentials?
You need to create an application on the Feishu Open Platform, enable Bot capabilities, and retrieve your App ID and App Secret. Enter these credentials along with your Anthropic API Key in EasyClaw's local setup to initialize the connection.
Does the skill stream outputs in real-time or send a single block?
The connection handler captures the terminal output buffer of Claude Code and streams it to Feishu in real-time chunk-by-chunk, allowing your team to watch the code generation and command execution as it happens.
Can I debug non-Kotlin and non-Python errors?
Yes. Claude Code is language-agnostic and can analyze, write, and debug code across all major languages — including TypeScript, JavaScript, Python, Go, Rust, Java, and C++, as well as HTML/CSS and SQL.
What is "workspace-scoped execution"?
Workspace-scoped execution means the AI's file-system permissions are locked to a specific project directory (e.g., `d:\\project\\easyclaw.com`). The bot cannot navigate outside this directory to read or edit other system files.
Can multiple developers interact with the bot simultaneously?
Yes. The message queue processes incoming queries sequentially, keeping separate execution states for different chat channels so multiple team discussions do not conflict.
Do I need a paid Anthropic API account?
Yes. Operating Claude Code requires a valid Anthropic API key with active credit balance to process language generation and code analysis requests.
Browse more in Development or all skills.
Add ClaudeIM to Your Workflow
Get EasyClaw, add this skill, and start building AI agent workflows in minutes.
Get EasyClaw Free →