AI导哈

系统日志搜索与错误模式记忆MCP服务

一个将系统日志搜索（sls）与错误模式记忆（slsm）整合为统一工具的MCP服务器，帮助AI编程智能体搜索日志、分析根因并复用已知修复方案。

  • by krasmussen37
  • Python
  • 本地部署
  • 更新于
在 GitHub 查看 krasmussen37/sls-slsm-mcp

使用说明

sls-slsm-mcp

MCP server bridging sls (System Log Search) and slsm (SLS Memory) into a single tool server for AI coding agents.

Tools

Tool Description
sls_search Search system logs, agent stderr, and application logs
sls_context Get surrounding log entries around a timestamp for root cause analysis
sls_timeline Activity timeline aggregated by time buckets
sls_alert Check error/warning thresholds (OK/WARNING/CRITICAL)
slsm_context Look up known fixes for an error from the pattern playbook
slsm_add_pattern Add a new error pattern with regex and metadata
slsm_feedback Mark a pattern as helpful or harmful
slsm_reflect Auto-extract new patterns from recent logs

Prerequisites

  • Python 3.7+
  • sls on PATH
  • slsm on PATH

Install

./install.sh                    # Interactive
./install.sh --configure-all    # Non-interactive, configure all agents
./install.sh --configure-claude # Claude Code only

Manual Configuration

Claude Code (~/.claude.json):

{
  "mcpServers": {
    "sls-slsm": {
      "command": "/home/you/.local/bin/sls-slsm-mcp",
      "args": [],
      "env": {},
      "description": "System log search (sls) + error pattern memory (slsm)"
    }
  }
}

Codex CLI (~/.codex/config.toml):

[mcp_servers.sls-slsm]
command = "/home/you/.local/bin/sls-slsm-mcp"
args = []

Gemini CLI (~/.gemini/settings.json):

{
  "mcpServers": {
    "sls-slsm": {
      "command": "/home/you/.local/bin/sls-slsm-mcp",
      "args": []
    }
  }
}

Environment Variables

Variable Default Description
SLS_BIN auto-detect Path to sls binary
SLSM_BIN auto-detect Path to slsm binary
SLS_SLSM_MCP_DEBUG unset Set to 1 to keep stderr open

Test

sls-slsm-mcp --version
echo '{"jsonrpc":"2.0","id":0,"method":"initialize","params":{}}' | sls-slsm-mcp

License

MIT

相关 MCP