MCP SSH Manager - SSH Remote Server Management via Model Context Protocol 🚀
A Model Context Protocol (MCP) server that enables Claude Code and OpenAI Codex to manage multiple SSH connections. Execute commands, transfer files, manage databases, create backups, monitor health, and automate DevOps tasks across your servers — directly from your AI assistant.
🎉 What's New in v3.4.0
Full Windows OpenSSH support + shell-agnostic session sync (Released: May 7, 2026)
🪟 Windows OpenSSH targets — encoding & syntax fixes (#31)
Mojibake fix — Commands on Windows hosts no longer inherit the OEM code page (CP950, CP932, CP1252…). Payloads are now wrapped as (the same approach used by Ansible / Chef / Puppet for Windows remote execution), sidestepping every quoting and code-page issue
⭐AI-driven public opinion & trend monitor with multi-platform aggregation, RSS, and smart alerts.🎯 告别信息过载，你的 AI 舆情监控助手与热点筛选工具！聚合多平台热点 + RSS 订阅，支持关键词精准筛选。AI 智能筛选新闻 + AI 翻译 + AI 分析简报直推手机，也支持接入 MCP 架构，赋能 AI 自然语言对话分析、情感洞察与趋势预测等。支持 Docker ，数据本地/云端自持。集成微信/飞书/钉钉/Telegram/邮件/ntfy/bark/slack 等渠道智能推送。
powershell -NoProfile -OutputFormat Text -EncodedCommand <utf16le-base64>
cmd.exe
cd && syntax fix — cd ${workingDir} && ${cmd} is invalid in cmd.exe; replaced with Set-Location '${escapedDir}'; ${cmd} (PowerShell-native, with ' → '' escaping) across ssh_execute, ssh_execute_group, and ssh_execute_sudo
Strictly gated behind platform = "windows" — Linux/macOS targets take the exact same code path as before, zero risk of regression
Thanks @WenKingSu for the clean, well-documented patch
The previous /[$#>]\s*$/ regex broke on custom prompts, ANSI color codes, multiline prompts, slow shells, .bashrc noise, AIX-style login flows — ssh_session_start would frequently fail with "Timeout waiting for shell prompt"
Now uses UUID v4 markers as protocol boundaries: each command is wrapped with a unique end marker (set +e; <cmd>; __mcp_status=$?; printf '\n<marker>:%s\n' "$__mcp_status"), and the PTY is opened with ECHO: 0 so the buffer contains only command output
Bonus: sessions now report real exit codes captured from $? instead of the previous !output.includes('command not found') heuristic
Resolves #20 and supersedes the proposed configurable prompt_pattern approach by making the problem disappear entirely — no per-server config required
Thanks @MakksSh for the precise root-cause analysis and the clean refactor
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
