alive-analysis — Claude Code MCP server: what it does, how to install
Tracks data investigations in Markdown files across a five-stage ALIVE loop. It keeps reasoning, hypotheses, and conclusions searchable in your repo.
claude mcp add --env ALIVE_ANALYSES_DIR=<ALIVE_ANALYSES_DIR> --transport stdio alive-analysis -- npx -y alive-analysis-mcpWhat alive-analysis does
The server turns one-off metric investigations into Git-tracked project documentation. Instead of losing context when a chat session ends, the agent walks through five steps: defining questions, inspecting data, investigating drivers, sharing conclusions, and tracking follow-ups. Every phase writes Markdown files directly into the analyses/ folder.
Three operation modes fit different workloads. Full mode splits the review across separate files with checklists, quick mode logs everything in a single document, and learn mode offers training scenarios. An internal engine routes tasks to up to 31 specialist agents depending on the stage.
Without the ALIVE_ANALYSES_DIR environment variable pointing to your target folder, the server cannot place the generated files. Keeping these records in version control prevents teams from re-running the same investigations months later.
What it helps with
- preserve data investigations and reasoning directly in Git history
- trace why metrics shifted without reconstructing the context from memory
- guide data checks through a structured five-stage evaluation gate
- query historical analysis files with full-text search commands
Example prompts for the agent
- /analysis-new
- /analysis-search "drop in checkout conversion"
- /analysis-status
How to install alive-analysis
Platform CLI
claude mcp add --env ALIVE_ANALYSES_DIR=<ALIVE_ANALYSES_DIR> --transport stdio alive-analysis -- npx -y alive-analysis-mcpReplace <…> with your own values
Platform CLI
codex mcp add alive-analysis --env ALIVE_ANALYSES_DIR=<ALIVE_ANALYSES_DIR> -- npx -y alive-analysis-mcpReplace <…> with your own values
Config file
{
"mcpServers": {
"alive-analysis": {
"command": "npx",
"args": [
"-y",
"alive-analysis-mcp"
],
"env": {
"ALIVE_ANALYSES_DIR": "<ALIVE_ANALYSES_DIR>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"alive-analysis": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"alive-analysis-mcp"
],
"env": {
"ALIVE_ANALYSES_DIR": "<ALIVE_ANALYSES_DIR>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e ALIVE_ANALYSES_DIR=<ALIVE_ANALYSES_DIR> alive-analysis npx -y alive-analysis-mcpReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about alive-analysis
How to install alive-analysis in Claude Code?
Run claude mcp add alive-analysis -- npx -y alive-analysis-mcp. If you want to specify a custom storage directory, run claude mcp add --env ALIVE_ANALYSES_DIR=<ALIVE_ANALYSES_DIR> --transport stdio alive-analysis -- npx -y alive-analysis-mcp.
What does alive-analysis do?
It enforces an analytical loop with five distinct stages, recording findings in Markdown files inside your repository. Future AI sessions can search these files to recall prior context and hypotheses.
What is the difference between Quick and Full modes?
Quick mode condenses the workflow into a single file for urgent checks. Full mode creates dedicated files for each phase with extended verification checklists and agent routing.
