[Feature Request] Add Support for webssh2-mcp to Enable LLM Agent Integration

[hxgh776]

Hi maintainers,

First, I want to express my gratitude for developing and maintaining webssh2—it has been an essential tool for web-based SSH access in my workflow, and its usability and reliability have been greatly appreciated.

I’m currently working on integrating webssh2 with LLM agents (e.g., AI-powered automation tools for SSH session management, intelligent command execution, or task orchestration via LLM prompts). During this process, I identified that webssh2-mcp (a component intended to extend webssh2’s capabilities for such integrations) is not yet supported by the current version of webssh2. This gap prevents LLM agents from establishing seamless, reliable communication with webssh2: without native support for webssh2-mcp, the agent cannot properly interact with webssh2 to initiate sessions, pass commands, or retrieve execution results—blocking the entire AI-augmented workflow.

Thus, I’m submitting this feature request to propose adding native support for webssh2-mcp in webssh2, with the core goal of enabling LLM agents to use webssh2 effectively. To make this integration practical, the support could focus on the following key areas:

1.Interface Compatibility: Ensure webssh2’s internal APIs/message protocols align with webssh2-mcp, enabling smooth data transmission (e.g., session initialization, command payloads, output feedback) between the two components.
2.Agent-Friendly Configuration: Add optional settings (e.g., webssh2-mcp endpoint specification, authentication bridging) to let LLM agents securely connect to webssh2 via webssh2-mcp without custom workarounds.
3.End-to-End Validation: Verify that the full chain (LLM agent → webssh2-mcp → webssh2 → target SSH server) works stably for common agent tasks (e.g., automated command runs, session persistence, error handling).

Supporting webssh2-mcp for LLM agents would expand webssh2’s utility into AI-augmented scenarios like DevOps automation, remote server troubleshooting, or batch task management—benefiting not just my use case, but likely other users looking to combine webssh2 with modern LLM capabilities.

[billchurch]

Believe it or not, this is a stretch goal for a release I'm working on. If it doesn't make it in, it'll be in a test branch at the very least.

Will reference this Issue when something is ready to test. Thanks for your submission!