Add OpenClaw environment type

[zkwentz]

Summary

Adds a new OpenClaw environment type following the MCP (Model Context Protocol) pattern.

Features

• 8 tools: exec, read, write, edit, web_search, web_fetch, memory_search, memory_get
• Full test suite (17 tests passing)
• Docker support via workflow integration
• Follows echo_env reference implementation pattern

Tools

Tool	Description
exec	Execute shell commands
read	Read file contents
write	Write files
edit	Edit files with search/replace
web_search	Search the web (simulated in sandbox)
web_fetch	Fetch URL content (simulated in sandbox)
memory_search	Semantic memory search
memory_get	Retrieve memory snippets

Testing

cd tests/envs
pytest test_openclaw_environment.py -v

Related

• OpenClaw: https://github.com/openclaw/openclaw
• MCP: https://modelcontextprotocol.io

[greptile-apps]

Greptile Summary

Adds a new OpenClaw environment following the MCP (Model Context Protocol) pattern established in RFC 003. The implementation provides 8 tools for agentic RL training: file operations (read/write/edit), shell execution (exec), web tools (web_search/web_fetch - simulated in sandbox), and memory management (memory_search/memory_get).

Key Changes:

• Environment server (openclaw_environment.py) extends MCPEnvironment with inline FastMCP tool definitions
• Client (client.py) is a simple wrapper of MCPToolClient for OpenClaw-specific usage
• Comprehensive test suite with 17 tests covering all tools, edge cases, and episode isolation
• Docker integration added to CI/CD workflow
• Per-episode workspace isolation using temp directories
• Follows echo_env reference implementation pattern

Architecture Alignment:

• Correctly implements dual API boundary (MCP tools for agents, Gym API for infrastructure)
• No client-server separation violations
• Agents cannot access reset/simulation controls (only exposed via infrastructure API)
• Shell execution uses shell=True which is acceptable given workspace sandboxing

Confidence Score: 5/5

• This PR is safe to merge with minimal risk
• Clean implementation following established patterns; no invariant violations; comprehensive test coverage; proper isolation and security controls; no Tier 1 issues found
• No files require special attention

Important Files Changed

Filename	Overview
envs/openclaw_env/server/openclaw_environment.py	Implements OpenClaw environment with 8 MCP tools; follows MCPEnvironment pattern correctly; shell execution uses shell=True which is acceptable in sandboxed context
envs/openclaw_env/client.py	Simple client wrapper extending MCPToolClient; no client-server separation violations; follows echo_env pattern
envs/openclaw_env/server/app.py	FastAPI app creation follows standard pattern; correctly uses create_app with MCP types
tests/envs/test_openclaw_environment.py	Comprehensive test suite with 17 tests covering all tools, edge cases, and episode isolation
envs/openclaw_env/server/Dockerfile	Proper Docker setup with system dependencies, workspace isolation, and health check

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Agent] -->|ListToolsAction| B[OpenClawEnvironment]
    A -->|CallToolAction| B
    
    B --> C{Tool Router}
    
    C -->|exec| D[Shell Execution]
    C -->|read/write/edit| E[File Operations]
    C -->|web_search/web_fetch| F[Web Tools - Simulated]
    C -->|memory_search/memory_get| G[Memory Management]
    
    D --> H[Workspace Sandbox]
    E --> H
    G --> H
    
    H -->|Observation| A
    F -->|Observation| A
    
    I[Infrastructure] -->|reset| B
    I -->|state| B
    
    B -->|Creates per episode| H
    
    style H fill:#e1f5ff
    style A fill:#fff4e1
    style I fill:#ffe1f5
    style B fill:#e1ffe1

Loading

_{Last reviewed commit: 984320c}