On-device Splunk investigation stack — deterministic Polars detectors exposed as MCP tools, with GitHub Copilot or Claude Code as the primary reasoning layer. Optional standalone LangGraph/Ollama agent for environments without Copilot/Claude. Traces errors to microservice call graphs. No data leaves the machine.
debaditya-mohankudo/splunk-intelligence
Skill Claude CodeCodex
Investigate a Splunk production issue. Loads events, runs deterministic detectors, then drives an iterative investigation loop via MCP tools — Claude is the reasoning engine. No Ollama, no API key, no server process. Watch live progress via the TUI (uv run python -m splunk.tui).
debaditya-mohankudo/splunk-intelligence
Instructions file CodexOpenCode
AGENTS.md instructions for debaditya-mohankudo/splunk-intelligence, covering splunk investigation agent — copilot instructions, mcp tools, before you start, investigation loop and step 1 — start.
debaditya-mohankudo/splunk-intelligence
Instructions file
Claude Code instructions for debaditya-mohankudo/splunk-intelligence, covering splunk intelligence — investigation stack, goal, project planning, project overview and stack.
debaditya-mohankudo/splunk-intelligence
MCP server Claude CodeCodexCursor +2
MCP server "splunk-analysis" as configured in debaditya-mohankudo/splunk-intelligence. Runs locally from the splunk-analysis Python package.