MicroClaw

https://microclaw.ai/docs
Admin
MicroClaw is a Rust multi-channel agent runtime for Telegram, Discord, Slack, Feishu/Lark, IRC, Web,...
Tokens:45,965
Snippets:639
Trust Score:4.7
Update:4 months ago
Show doc for...
Tokens:
Raw
### Quick Start: Clone and Run Microclaw

Source: https://microclaw.ai/docs/developer-guide

Clone the repository, navigate to the directory, configure credentials, and start the application.

```bash
git clone <repo-url>  
cd microclaw  
cp microclaw.config.example.yaml microclaw.config.yaml  
# Edit microclaw.config.yaml with your credentials  
cargo run -- start  

```

--------------------------------

### Setup Microclaw Configuration (Interactive)

Source: https://microclaw.ai/docs/installation

Use the interactive setup wizard to configure Microclaw, which includes provider and model selection.

```bash
microclaw setup  
```

--------------------------------

### Install MicroClaw with One-line Installer

Source: https://microclaw.ai/docs/quickstart

Use this command to install prebuilt GitHub release binaries for MicroClaw.

```bash
curl -fsSL https://microclaw.ai/install.sh | bash  

```

--------------------------------

### Run Interactive Microclaw Setup

Source: https://microclaw.ai/docs/quickstart

Use the interactive setup wizard to configure Microclaw. It validates settings, tests connectivity, and generates a `microclaw.config.yaml` file.

```bash
microclaw setup  

```

--------------------------------

### Install MicroClaw Gateway Service

Source: https://microclaw.ai/docs/quickstart

Install the MicroClaw gateway service using the gateway install command. The --force flag can be used to overwrite an existing installation.

```bash
microclaw gateway install  
microclaw gateway install --force  

```

--------------------------------

### Install MicroClaw Agent Runtime on macOS

Source: https://microclaw.ai

Recommended installation script for most macOS setups. Executes a shell script to install the MicroClaw agent runtime.

```bash
curl -fsSL https://microclaw.ai/install.sh | bash
```

--------------------------------

### Enable Sandbox Mode

Source: https://microclaw.ai/docs/quickstart

Quick commands to enable sandbox mode and verify its setup.

```bash
microclaw setup --enable-sandbox
microclaw doctor sandbox
```

--------------------------------

### Example Plugin Manifest

Source: https://microclaw.ai/docs/plugins

A comprehensive example of a plugin manifest defining commands, tools, and context providers with their respective configurations and permissions.

```yaml
name: ops
enabled: true

commands:
  - command: /uptime
    description: Show host uptime
    run:
      command: "uptime"
      timeout_secs: 10
      execution_policy: host_only

  - command: /safe-ls
    description: List current chat working directory in sandbox
    run:
      command: "ls -la"
      timeout_secs: 10
      execution_policy: sandbox_only

  - command: /announce
    description: Echo command args
    response: "Announcement: {{args}}"

tools:
  - name: plugin_safe_ls
    description: List files in the plugin working directory
    input_schema:
      type: object
      properties: {}
      required: []
    permissions:
      execution_policy: sandbox_only
      allowed_channels: ["telegram", "discord", "web"]
    run:
      command: "ls -la"
      timeout_secs: 10

context_providers:
  - name: policy_prompt
    kind: prompt
    content: "Always include a short risk summary for shell actions in channel {{channel}}."

  - name: runbook_doc
    kind: document
    permissions:
      execution_policy: host_only
    run:
      command: "cat ./docs/operations/runbook.md"
      timeout_secs: 10

```

--------------------------------

### Deployment Check Workflow

Source: https://microclaw.ai/docs/usage

This example illustrates a workflow for checking system status, including port usage and configuration file analysis.

```plaintext
You: Check if port 8080 is in use, read the nginx config, and tell me if everything looks correct  
Bot: [bash: lsof -i :8080 -> read_file: /etc/nginx/nginx.conf -> analysis]  

```

--------------------------------

### Install and Manage MicroClaw Gateway Service

Source: https://microclaw.ai/docs/installation

Commands to install, manage the status, start, stop, restart, and view logs for the persistent gateway service. Use --force to reinstall.

```bash
microclaw gateway install
```

```bash
microclaw gateway install --force
```

```bash
microclaw gateway status
```

```bash
microclaw gateway status --json --deep
```

```bash
microclaw gateway start
```

```bash
microclaw gateway stop
```

```bash
microclaw gateway restart
```

```bash
microclaw gateway logs 200
```

```bash
microclaw gateway uninstall
```

--------------------------------

### Start MicroClaw Service

Source: https://microclaw.ai/docs/channel-setup

Use this command to start the MicroClaw service. Access the Web UI via the default local address provided in the logs.

```bash
microclaw start  

```

--------------------------------

### Install MicroClaw (Full Variant)

Source: https://microclaw.ai/docs/installation

This command installs the full variant of MicroClaw, which includes support for the Matrix channel. It uses the same one-line installer script with an additional flag.

```bash
curl -fsSL https://microclaw.ai/install.sh | bash -s -- --full
```

--------------------------------

### Install MicroClaw with Homebrew on macOS

Source: https://microclaw.ai/docs/quickstart

Install MicroClaw using Homebrew by tapping the microclaw repository and installing the package.

```bash
brew tap microclaw/tap  
brew install microclaw  

```

--------------------------------

### Install MicroClaw on Windows (Full Variant - PowerShell)

Source: https://microclaw.ai/docs/installation

Install the full variant of MicroClaw on Windows, including Matrix channel support, using PowerShell. This command downloads and executes the installation script with the -Full flag.

```powershell
& ([scriptblock]::Create((iwr https://microclaw.ai/install.ps1 -UseBasicParsing).Content)) -Full
```

--------------------------------

### Adapt Skill Prompt Example

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Use this prompt to find a skill for a specific task, such as converting docs to PDF, and adapt it for your local repository.

```bash
Find a skill for converting docs to PDF and adapt it to this repo.  

```

--------------------------------

### Start MicroClaw MCP Server

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Starts the MicroClaw MCP server with informational logging enabled. Monitor the logs for connection messages to verify server startup.

```bash
RUST_LOG=info cargo run -- start
```

--------------------------------

### Execute Shell Command (List Files)

Source: https://microclaw.ai/docs/usage

This example demonstrates running the 'ls' command to list files in the current directory via the bot.

```text
You: List all files in the current directory
Bot: [runs ls, returns file listing]
```

--------------------------------

### Install MicroClaw on Windows (PowerShell)

Source: https://microclaw.ai/docs/installation

Install MicroClaw on Windows using PowerShell. This command downloads and executes the installation script.

```powershell
iwr https://microclaw.ai/install.ps1 -UseBasicParsing | iex
```

--------------------------------

### Install Microclaw Binary

Source: https://microclaw.ai/docs/installation

Copy the compiled release binary to a directory on your system's PATH for global access.

```bash
cp target/release/microclaw /usr/local/bin/  
```

--------------------------------

### Install OpenSSL 3 Runtime on Debian/Ubuntu

Source: https://microclaw.ai/docs/installation

If your system has a compatible glibc version but is missing OpenSSL 3, install the necessary runtime package using apt.

```bash
apt install libssl3
```

--------------------------------

### Install Microclaw with Homebrew (macOS)

Source: https://microclaw.ai/docs/installation

Installs the standard Microclaw package using Homebrew. Ensure you have Homebrew installed.

```bash
brew tap microclaw/tap
brew install microclaw
```

--------------------------------

### SKILL.md Frontmatter Example

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Define your skill's metadata using the SKILL.md file. Supported fields include name, description, platforms, and dependencies.

```yaml
---
name: my-skill
description: One-line summary of what the skill does.
platforms: [darwin, linux, windows]
deps: [curl]
---

# My Skill

When to use:
- ...

Workflow:
1. ...
2. ...

```

--------------------------------

### Start Streaming Chat Run

Source: https://microclaw.ai/docs/web-operator-api

Initiates an asynchronous chat run. Use this to start a conversation and get a run_id for tracking progress.

```json
{
  "session_key": "ops-bot",
  "sender_name": "automation",
  "message": "Summarize the latest incidents"
}

```

--------------------------------

### Run Microclaw Docker Container (Quick Start)

Source: https://microclaw.ai/docs/installation

Pulls the latest Microclaw Docker image and runs it. This is a quick way to try Microclaw without persistent storage or custom configuration.

```bash
docker pull ghcr.io/microclaw/microclaw:latest
docker run --rm -it \
  -p 127.0.0.1:10961:10961 \
  ghcr.io/microclaw/microclaw:latest
```

--------------------------------

### Read Memory Example

Source: https://microclaw.ai/docs/usage

Shows how to retrieve information stored in memory, including personal details and global memory content.

```text
You: What do you remember about me?  
Bot: [reads memory and responds with stored info]  


```

```text
You: Show me your global memory  
Bot: [reads and displays AGENTS.md content]  


```

--------------------------------

### Install apple-calendar dependencies

Source: https://microclaw.ai/docs/skills

Install 'ical-buddy' for querying Apple Calendar events. 'osascript' is built-in on macOS.

```bash
brew install ical-buddy
```

--------------------------------

### Manual Microclaw Configuration Example

Source: https://microclaw.ai/docs/quickstart

Manually configure Microclaw by editing the `microclaw.config.yaml` file. This example shows settings for LLM provider, API key, model, data directories, timezone, and Telegram channel integration.

```yaml
llm_provider: "anthropic"  
api_key: "sk-ant-..."  
model: "claude-sonnet-4-5-20250929"  
data_dir: "~/.microclaw"  
working_dir: "~/.microclaw/working_dir"  
override_timezone: "Asia/Shanghai" # optional; default uses system timezone when unset  
working_dir_isolation: "chat" # optional; defaults to "chat"  
max_document_size_mb: 100  
  
# Pick one or more channels:  
channels:  
  telegram:  
    default_account: "main"  
    # optional channel-level provider profile override  
    # provider_preset: "ops-openrouter"  
    # optional: route Telegram topics as isolated chats ("<chat_id>:<thread_id>")  
    # topic_routing:  
    #   enabled: true  
    accounts:  
      main:  
        bot_token: "123456:ABC-DEF1234..."  
        bot_username: "my_bot"  
      support:  
        bot_token: "987654:XYZ-DEF9999..."  
        bot_username: "support_bot"  
        # optional per-account provider profile override  
        # provider_preset: "deepseek-ops"  
        # optional per-account topic routing override  
        # topic_routing:  
        #   enabled: false  
# Telegram group chat note for multi-bot:  
#   In BotFather for each bot, enable Allow Groups and disable Group Privacy.  
# Other channels:  
# channels:  
#   discord:  
#     accounts:  
#       main: { bot_token: "DISCORD_TOKEN_MAIN" }  
#   slack:  
#     accounts:  
#       main: { bot_token: "xoxb-main...", app_token: "xapp-main..." }  
#   feishu:  
#     accounts:  
#       main: { app_id: "cli_xxx", app_secret: "xxx", topic_mode: true } # topic_mode only for feishu/lark domains  
# Optional reusable provider profiles:  
# provider_presets:  
#   ops-openrouter:  
#     provider: "openrouter"  
#     api_key: "sk-or-..."  
#     default_model: "openai/gpt-4o-mini"  
#   deepseek-ops:  
#     provider: "deepseek"  
#     api_key: "sk-ds-..."  
#     default_model: "deepseek-chat"  
# channels:  
#   irc:  
#     server: "irc.example.com"  
#     nick: "microclaw"  
#     channels: "#general"  
web_enabled: true  

```

--------------------------------

### Install OpenSSL 3 Runtime on RHEL-family

Source: https://microclaw.ai/docs/installation

If your system has a compatible glibc version but is missing OpenSSL 3, install the necessary runtime package using dnf.

```bash
dnf install openssl3-libs
```

--------------------------------

### Sync Skills from Registry Tool Input

Source: https://microclaw.ai/docs/skills

Example input for the 'sync_skills' tool to import remote skills and normalize their metadata.

```json
{
  "skill_name": "find-skills",
  "source_repo": "vercel-labs/skills",
  "git_ref": "main",
  "target_name": "find-skills"
}
```

--------------------------------

### Write Memory Example

Source: https://microclaw.ai/docs/usage

Demonstrates how to save information to the bot's memory. This includes personal preferences and specific facts like server IPs.

```text
You: Remember that my preferred programming language is Rust  
Bot: Noted. Saved memory #123: my preferred programming language is Rust  


```

```text
You: Save to global memory: The production server IP is 10.0.1.50  
Bot: Memory saved to global scope.  


```

--------------------------------

### Example SOUL.md Configuration Path

Source: https://microclaw.ai/docs/usage

This configuration snippet shows how to explicitly set the path to the SOUL.md file. If not provided, MicroClaw auto-discovers it.

```yaml
# Optional: explicit path (otherwise auto-discovered)  
soul_path: "./SOUL.md"  


```

--------------------------------

### Web Search Example

Source: https://microclaw.ai/docs/usage

Demonstrates how to perform a web search using the 'Search the web' command. The bot returns titles, URLs, and snippets for the top results.

```text
You: Search the web for "Rust async programming best practices"  
Bot: 1. Title...  
     URL...  
     Snippet...  
  
     2. Title...  
     ...  


```

```text
You: What's the latest news about Anthropic?  
Bot: [searches and summarizes results]  


```

--------------------------------

### Web Fetch Example

Source: https://microclaw.ai/docs/usage

Shows how to fetch content from a URL. The bot returns plain text after stripping HTML tags, with a maximum limit of 20KB.

```text
You: Fetch https://example.com and tell me what it says  
Bot: [fetches page, strips HTML, returns text content]  


```

```text
You: Go to https://httpbin.org/get and show me the response  
Bot: [fetches and displays the JSON response]  


```

--------------------------------

### Install apple-notes dependency

Source: https://microclaw.ai/docs/skills

Install the 'memo' tool required for managing Apple Notes on macOS.

```bash
brew tap antoniorodr/memo
brew install antoniorodr/memo/memo
```

--------------------------------

### Example MCP Namespaced Tool

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Illustrates the naming convention for MCP namespaced tools. Tools are prefixed with 'mcp_<server_name>_<tool_name>'.

```bash
mcp_filesystem_read_file
```

--------------------------------

### Install apple-reminders dependency

Source: https://microclaw.ai/docs/skills

Install the 'remindctl' tool required for managing Apple Reminders on macOS.

```bash
brew install steipete/tap/remindctl
```

--------------------------------

### Microclaw Configuration File Example

Source: https://microclaw.ai/docs/installation

Manually create or modify the `microclaw.config.yaml` file to define LLM providers, API keys, models, data directories, and enabled channels.

```yaml
llm_provider: "anthropic"  
api_key: "..."  
model: "claude-sonnet-4-5-20250929"  
data_dir: "~/.microclaw"  
working_dir: "~/.microclaw/working_dir"  
max_document_size_mb: 100  
  
# Enable at least one channel:  
channels:  
  telegram:  
    default_account: "main"  
    accounts:  
      main:  
        bot_token: "..."  
        bot_username: "..."  
# Optional reusable provider profiles:  
# provider_presets:  
#   ops-openrouter:  
#     provider: "openrouter"  
#     api_key: "sk-or-..."  
#     default_model: "openai/gpt-4o-mini"  
# Optional additional channels:  
# channels:  
#   discord:  
#     accounts:  
#       main: { bot_token: "..." }  
#   slack:  
#     accounts:  
#       main: { bot_token: "...", app_token: "..." }  
#   feishu:  
#     accounts:  
#       main: { app_id: "...", app_secret: "..." }  
#   irc:  
#     server: "irc.example.com"  
#     nick: "microclaw"  
#     channels: "#general"  
web_enabled: true  


```

--------------------------------

### Run Microclaw from Source (Development)

Source: https://microclaw.ai/docs/installation

Execute the Microclaw application directly from the source code in development mode, starting the scheduler and configured adapters.

```bash
cargo run -- start  
```

--------------------------------

### Install Microclaw Full Variant with Homebrew (macOS)

Source: https://microclaw.ai/docs/installation

Installs the full Microclaw variant, which includes Matrix channel support. Use this if you need extended features.

```bash
brew install microclaw-full
```

--------------------------------

### Query and Create Apple Calendar Events

Source: https://microclaw.ai/docs/skills

Examples for querying upcoming events with 'icalBuddy' and creating new events using 'osascript'.

```bash
icalBuddy eventsToday
icalBuddy eventsFrom:today to:7 days from now
osascript -e 'tell application "Calendar" to tell calendar "Work" to make new event with properties {summary:"Team Sync", start date:date "Monday, February 10, 2026 10:00:00", end date:date "Monday, February 10, 2026 10:30:00"}'
```

--------------------------------

### Execute Shell Command (System Info)

Source: https://microclaw.ai/docs/usage

Run any bash command on the host machine and receive its standard output, standard error, and exit code. This example shows retrieving system information.

```text
You: Run `uname -a` for me
Bot: [executes bash, returns system info]
```

--------------------------------

### Manage Apple Notes

Source: https://microclaw.ai/docs/skills

Examples for interacting with Apple Notes using the 'memo' tool. Use 'memo notes' to list, search, add, or edit notes.

```bash
memo notes
memo notes -s "weekly plan"
memo notes -a "Project Ideas"
memo notes -e
```

--------------------------------

### Find Skills Prompt Example

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Use this prompt to search for skills related to a specific task, like 'playwright test debugging', from the vercel-labs/skills repository.

```bash
Find skills for playwright test debugging from vercel-labs/skills.  

```

--------------------------------

### Example of Semantic Memory Interaction

Source: https://microclaw.ai/docs/memory

Illustrates a user preference being remembered and later extracted by the system, demonstrating the flow of semantic memory.

```text
You: Remember that I prefer Rust examples  
Bot: Noted. Saved memory #123: I prefer Rust examples.  
  
[15 minutes later, automatically]  
Reflector: extracted "User prefers Rust code examples" → memories table  
  
[Next session after /reset]  
Bot: [has both AGENTS.md + structured memories in context]  

```

--------------------------------

### Manage Structured Memories

Source: https://microclaw.ai/docs/usage

Examples for searching, updating, and deleting specific memory entries using their IDs. This allows for precise control over stored information.

```text
You: Search memory for "production database"  
Bot: [id=42] [KNOWLEDGE] [chat] production database is on port 5433  


```

```text
You: Update memory id 42 to "production database is on port 6432"  
Bot: Memory id=42 updated.  


```

```text
You: Delete memory id 42  
Bot: Memory id=42 archived.  


```

--------------------------------

### Configure ACP Subagent Runtime

Source: https://microclaw.ai/docs/usage

Example configuration for spawning subagents on an ACP-backed external worker. Specify the runtime and optionally the runtime target.

```json
{
  "task": "Review the repo and summarize risky files",
  "runtime": "acp",
  "runtime_target": "reviewer"
}


```

--------------------------------

### Add Sidecar Integration Configuration

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Copies an example configuration file for a HAPI bridge sidecar integration. This is typically placed in the MCP configuration directory.

```bash
cp mcp.hapi-bridge.example.json ~/.microclaw/mcp.d/hapi-bridge.json
```

--------------------------------

### Manage Apple Reminders

Source: https://microclaw.ai/docs/skills

Examples for interacting with Apple Reminders using the 'remindctl' tool. Use commands to list, add, complete, or manage reminder lists.

```bash
remindctl today
remindctl add "Buy milk"
remindctl complete 1
remindctl list
```

--------------------------------

### Minimal MCP Configuration

Source: https://microclaw.ai/docs/memory

This is a minimal configuration example for the MCP memory backend. It specifies the default protocol version and the details for the 'memory' MCP server, including transport, endpoint, and authentication headers.

```json
{
  "defaultProtocolVersion": "2024-11-05",
  "mcpServers": {
    "memory": {
      "transport": "streamable_http",
      "endpoint": "http://127.0.0.1:8090/mcp",
      "headers": {
        "Authorization": "Bearer REPLACE_ME"
      },
      "request_timeout_secs": 60
    }
  }
}

```

--------------------------------

### Configure Provider Presets and Channel Overrides

Source: https://microclaw.ai/docs/configuration

Define reusable provider presets and apply them to channels and accounts. This example shows setting up 'ops-openrouter' and 'deepseek-ops' presets and assigning them to Telegram channels and accounts, including model-specific overrides.

```yaml
llm_provider: "anthropic"
api_key: "sk-ant-..." # global main profile
model: "claude-sonnet-4-5-20250929"

provider_presets:
  ops-openrouter:
    provider: "openrouter"
    api_key: "sk-or-..."
    llm_base_url: "https://openrouter.ai/api/v1" # optional; defaults from provider matrix
    llm_user_agent: "microclaw-ops/1.0" # optional
    default_model: "openai/gpt-4o-mini"
    show_thinking: false
  deepseek-ops:
    provider: "deepseek"
    api_key: "sk-ds-..."
    default_model: "deepseek/deepseek-chat"

channels:
  telegram:
    provider_preset: "ops-openrouter"
    default_account: "main"
    accounts:
      main:
        enabled: true
        bot_token: "xxx"
      ops:
        enabled: true
        bot_token: "yyy"
        provider_preset: "deepseek-ops"
        model: "deepseek/deepseek-chat"

openai_compat_body_overrides:
  temperature: 0.2

openai_compat_body_overrides_by_model:
  "openai/gpt-4o-mini":
    temperature: 0.0
  "deepseek/deepseek-chat":
    top_p: null
    reasoning_effort: "high"

```

--------------------------------

### Manage Scheduled Tasks

Source: https://microclaw.ai/docs/scheduler

Examples of natural language commands to manage scheduled tasks, including listing, pausing, resuming, and canceling tasks.

```natural language
"List my scheduled tasks"  
```

```natural language
"Pause task #3"  
```

```natural language
"Resume task #3"  
```

```natural language
"Cancel task #3"  
```

--------------------------------

### Build MicroClaw from Source

Source: https://microclaw.ai/docs/quickstart

Clone the MicroClaw repository, build the release binary, and copy it to your system's PATH.

```bash
git clone https://github.com/microclaw/microclaw.git  
cd microclaw  
cargo build --release  
cp target/release/microclaw /usr/local/bin/  

```

--------------------------------

### Initialize MCP Configuration

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Copies the minimal MCP configuration and creates the configuration directory. This sets up a basic local MCP server using filesystem transport.

```bash
cp mcp.minimal.example.json ~/.microclaw/mcp.json
mkdir -p ~/.microclaw/mcp.d
```

--------------------------------

### Execute Shell Command (Sequential Output)

Source: https://microclaw.ai/docs/usage

Demonstrates running a sequence of shell commands, including a delay, and capturing their combined output.

```text
You: Run `echo hello && sleep 2 && echo done`
Bot: hello
     done
```

--------------------------------

### Cron Expression Example: Every 5 Minutes

Source: https://microclaw.ai/docs/scheduler

This example demonstrates a cron expression that schedules a task to run every 5 minutes.

```cron
0 */5 * * * *  
```

--------------------------------

### List Plugins

Source: https://microclaw.ai/docs/usage

Use the /plugins list command to view available plugins for the bot.

```text
/plugins list
```

--------------------------------

### Install Browser Automation Dependency

Source: https://microclaw.ai/docs/installation

Install the 'agent-browser' npm package globally. This is an optional dependency required for the 'browser' tool. MicroClaw will use 'agent-browser.cmd' on Windows if available.

```bash
npm install -g agent-browser
```

```bash
agent-browser install
```

--------------------------------

### Validate Plugins

Source: https://microclaw.ai/docs/usage

Use the /plugins validate command to check the integrity and configuration of installed plugins.

```text
/plugins validate
```

--------------------------------

### Streaming Chat API Response

Source: https://microclaw.ai/docs/web-operator-api

The expected response shape after starting a streaming chat run.

```json
{
  "ok": true,
  "run_id": "6f4c2b1d-...".
  "session_key": "ops-bot",
  "chat_id": 123
}

```

--------------------------------

### Mission Control WebSocket chat.send Request

Source: https://microclaw.ai/docs/web-operator-api

Example of a chat.send request frame for sending a message through the WebSocket bridge.

```json
{
  "type": "req",
  "id": "send-1",
  "method": "chat.send",
  "params": {
    "sessionKey": "ops-bot",
    "message": "Summarize the latest incidents",
    "idempotencyKey": "idem-1"
  }
}

```

--------------------------------

### Uninstall MicroClaw via Homebrew

Source: https://microclaw.ai/docs/installation

Commands to uninstall MicroClaw if it was installed using Homebrew, including untapping the MicroClaw repository.

```bash
brew uninstall microclaw
```

```bash
brew untap microclaw/tap
```

--------------------------------

### Config Self-check

Source: https://microclaw.ai/docs/web-operator-api

Retrieves startup risk warnings and the overall risk level to identify unsafe settings.

```APIDOC
## GET /api/config/self_check

### Description
Returns startup risk warnings and `risk_level` (`none` / `medium` / `high`) to quickly spot unsafe settings. Current checks include sandbox posture and runtime availability, high-risk tool confirmation posture, ACP subagent runtime issues (e.g., bad `default_target`, missing worker commands, enabled `auto_approve`), web fetch validation posture, and web request-rate settings that drift above typical safe defaults.

### Method
GET

### Endpoint
/api/config/self_check

### Response
#### Success Response (200)
- **risk_level** (string) - Indicates the overall risk level: `none`, `medium`, or `high`.
- **warnings** (array) - A list of specific risk warnings identified during the self-check.
```

--------------------------------

### Mission Control WebSocket sessions_spawn Request

Source: https://microclaw.ai/docs/web-operator-api

Example of a sessions_spawn request frame for creating a new session via the WebSocket bridge.

```json
{
  "type": "req",
  "id": "spawn-1",
  "method": "sessions_spawn",
  "params": {
    "task": "Summarize active incidents",
    "label": "Incidents"
  }
}

```

--------------------------------

### Microclaw CLI Help

Source: https://microclaw.ai/docs/installation

Display CLI usage information for Microclaw.

```bash
microclaw help  
```

--------------------------------

### Mission Control WebSocket chat Event

Source: https://microclaw.ai/docs/web-operator-api

Example of a chat event received from the WebSocket bridge, indicating the final state of a message.

```json
{
  "type": "event",
  "event": "chat",
  "payload": {
    "runId": "6f4c2b1d-...".
    "sessionKey": "ops-bot",
    "seq": 1,
    "state": "final",
    "message": {
      "role": "assistant",
      "content": [{ "type": "text", "text": "..." }]
    }
  }
}

```

--------------------------------

### Mission Control WebSocket Connect Frame

Source: https://microclaw.ai/docs/web-operator-api

Example of a connect frame sent to establish a WebSocket connection with the Mission Control bridge.

```json
{
  "type": "req",
  "id": "connect-1",
  "method": "connect",
  "params": {
    "minProtocol": 3,
    "maxProtocol": 3,
    "auth": { "token": "mc_..." }
  }
}

```

--------------------------------

### Discover and Map Skills from Registry

Source: https://microclaw.ai/docs/skills

Use 'curl' to fetch skill information from a remote repository or search GitHub for specific skills. Useful for discovering and integrating external skills.

```bash
curl -sL "https://raw.githubusercontent.com/vercel-labs/skills/main/README.md"
curl -s "https://api.github.com/search/code?q=repo:vercel-labs/skills+playwright"
```

--------------------------------

### List Skills Command

Source: https://microclaw.ai/docs/skills-mcp-tutorial

Use the `/skills` command in the chat interface to list all available skills after restarting the service.

```bash
/skills  

```

--------------------------------

### Mission Control WebSocket session_setLabel Request

Source: https://microclaw.ai/docs/web-operator-api

Example of a session_setLabel request frame for updating a session's label via the WebSocket bridge.

```json
{
  "type": "req",
  "id": "label-1",
  "method": "session_setLabel",
  "params": {
    "sessionKey": "ops-bot",
    "label": "Ops"
  }
}

```

--------------------------------

### Microclaw Run Help

Source: https://microclaw.ai/docs/installation

Display CLI usage information for Microclaw when run via cargo.

```bash
cargo run -- help  
```

--------------------------------

### High-Level Runtime Flow Diagram

Source: https://microclaw.ai/docs/architecture

Illustrates the data flow from channel events through the gateway adapter to the agent engine, interacting with LLMs and tools, and persisting state.

```text
Channel event -> Gateway adapter -> Agent engine -> LLM + Tools loop  
                 |                                 |
                 v                                 v
           channel persistence               state + memory + sessions  

```

--------------------------------

### GET / (WebSocket Bridge)

Source: https://microclaw.ai/docs/web-operator-api

Provides a Mission Control-compatible WebSocket bridge for real-time interaction, including chat, session management, and agent/model information.

```APIDOC
## GET / (WebSocket Bridge)

### Description
Exposes an OpenClaw-style WebSocket bridge at the root path (`/`). This bridge is intended for Mission Control-style operators and supports various real-time interactions including chat, session lifecycle controls, and configuration retrieval.

### Method
GET

### Endpoint
/

### Notes
- The bridge is activated via a WebSocket upgrade request.
- It expects a `connect.challenge` event on socket open and a `connect` request frame.
- Supports `chat.send`, `chat.history`, `session_delete`, `sessions_send`, `sessions_kill`, `sessions_spawn`, and various `session_set*` methods.
- Advertised methods include `health`, `status`, `agents.list`, `models.list`, `config.get`, and `node.list`.

### Request Example (Connect Frame)
```json
{
  "type": "req",
  "id": "connect-1",
  "method": "connect",
  "params": {
    "minProtocol": 3,
    "maxProtocol": 3,
    "auth": { "token": "mc_..." }
  }
}
```

### Request Example (chat.send)
```json
{
  "type": "req",
  "id": "send-1",
  "method": "chat.send",
  "params": {
    "sessionKey": "ops-bot",
    "message": "Summarize the latest incidents",
    "idempotencyKey": "idem-1"
  }
}
```

### Response Example (chat event)
```json
{
  "type": "event",
  "event": "chat",
  "payload": {
    "runId": "6f4c2b1d-...",
    "sessionKey": "ops-bot",
    "seq": 1,
    "state": "final",
    "message": {
      "role": "assistant",
      "content": [{ "type": "text", "text": "..." }]
    }
  }
}
```
```

--------------------------------

### Run Preflight Diagnostics

Source: https://microclaw.ai/docs/quickstart

Execute the doctor command to perform preflight diagnostics on your MicroClaw installation. This helps identify potential issues before running the service.

```bash
microclaw doctor  

```

--------------------------------

### Streaming Web API: Initiate Async Run

Source: https://microclaw.ai/docs/testing

Initiates an asynchronous run using the Web API. Ensure you have your API key set in the MICROCLAW_API_KEY environment variable.

```bash
curl -sS http://127.0.0.1:10961/api/send_stream \
  -H "Authorization: Bearer $MICROCLAW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"session_key":"ops-bot","sender_name":"automation","message":"summarize the current repo"}'  

```

--------------------------------

### read_file

Source: https://microclaw.ai/docs/tools

Reads the content of a file, optionally starting from a specified line number and limiting the number of lines read. Output includes line numbers.

```APIDOC
## read_file

### Description
Read file content with line numbers.

### Parameters
#### Path Parameters
- **path** (string) - Required - File path to read
- **offset** (integer) - Optional - Line number to start from (1-based)
- **limit** (integer) - Optional - Maximum lines to read (default: 2000)

### Output format
`{line_number}\t{content}` per line.
```

--------------------------------

### File Write and Read Test

Source: https://microclaw.ai/docs/testing

Verify the bot's ability to create a file with specified content and then read its content. Ensure file paths and content are handled correctly.

```text
You: Create a file /tmp/microclaw_test.txt with the content "MicroClaw test file"  
```

```text
You: Read the file /tmp/microclaw_test.txt  
```

--------------------------------

### POST /api/send_stream

Source: https://microclaw.ai/docs/web-operator-api

Starts an asynchronous chat run. It takes a message and returns a run_id to consume progress and output via SSE from the /api/stream endpoint.

```APIDOC
## POST /api/send_stream

### Description
Starts an asynchronous chat run and returns a `run_id`. Progress, tool events, deltas, and final output can be consumed using Server-Sent Events (SSE) from the `GET /api/stream` endpoint.

### Method
POST

### Endpoint
/api/send_stream

### Parameters
#### Request Body
- **session_key** (string) - Required - The key for the chat session.
- **sender_name** (string) - Required - The name of the sender.
- **message** (string) - Required - The message to send.

### Request Example
```json
{
  "session_key": "ops-bot",
  "sender_name": "automation",
  "message": "Summarize the latest incidents"
}
```

### Response
#### Success Response (200)
- **ok** (boolean) - Indicates if the request was successful.
- **run_id** (string) - The ID of the asynchronous run.
- **session_key** (string) - The key of the chat session.
- **chat_id** (integer) - The ID of the chat.

#### Response Example
```json
{
  "ok": true,
  "run_id": "6f4c2b1d-...",
  "session_key": "ops-bot",
  "chat_id": 123
}
```
```

--------------------------------

### Run Microclaw Docker Container (Recommended)

Source: https://microclaw.ai/docs/installation

Runs the Microclaw Docker container with mounted configuration and persistent data directories. This is recommended for day-to-day use to maintain state and custom settings.

```bash
mkdir -p data tmp
chmod a+r microclaw.config.yaml
chmod -R a+rwX data tmp

docker run --rm -it \
  -p 127.0.0.1:10961:10961 \
  -v "$(pwd)/microclaw.config.yaml:/app/microclaw.config.yaml:ro" \
  -v "$(pwd)/data:/home/microclaw/.microclaw" \
  -v "$(pwd)/tmp:/app/tmp" \
  ghcr.io/microclaw/microclaw:latest
```

--------------------------------

### Mid-Conversation Messaging Example

Source: https://microclaw.ai/docs/usage

LLM can send messages to the chat during tool execution for progress updates on long tasks. This is used automatically for multi-step tasks.

```text
You: Analyze all log files in /var/log and give me a summary  
Bot: [sends "Scanning log files..." as a progress update]  
Bot: [sends "Found 15 log files, analyzing..." as another update]  
Bot: [final summary response]  
```

--------------------------------

### Build Microclaw with Matrix Support

Source: https://microclaw.ai/docs/installation

Build the release version of Microclaw with the 'full' feature enabled to include Matrix channel support.

```bash
cargo build --release --features full  
```

--------------------------------

### Enable Container Sandbox via CLI

Source: https://microclaw.ai/docs/configuration

Quickly enable the container sandbox feature using the MicroClaw CLI. Also includes a command to check sandbox status.

```bash
microclaw setup --enable-sandbox
microclaw doctor sandbox

```
Tokens:
Raw