life. scored. is just math over public rules — so any AI or agent can run it for you. Point your assistant at the resources below: it reads the rulebook, asks you for whatever it doesn't know, and computes your score on its own side. Your answers never reach us — the same promise as the website. We host the recipe, not the kitchen.
Drop this into any assistant that can fetch a URL (Claude, ChatGPT with browsing, Claude Code, …):
Fetch https://lifescored.com/llms.txt and read it, then fetch https://lifescored.com/rules.json for the exact rules, weights, formulas, input schema, and engine constants. Compute my life score: ask me for the inputs in the schema (only the ones you can't infer), then for each rule compute points = round(clamp(position, bounds) × weight) using the formulas in the file, sum them for my composite, and show the breakdown by rule with the one-line reasoning. Do all the math yourself — do not send my answers to any server. Be ready to explain how any rule works and what's deliberately left out.
MCP-capable agents can add the endpoint as a tool server (e.g. claude mcp add --transport http lifescored https://lifescored.com/mcp), or test it with npx @modelcontextprotocol/inspector.
Nothing about a person should ever be sent to lifescored.com to get a score, and nothing is. The MCP server computes nothing and accepts no personal data — its tools only hand back the public rules. All scoring happens on your device, or in your agent.
The weights are editorial and the rules are arguable — that's the design. If your agent (or you) thinks a weight is off, a source is stale, or a rule is missing, that feedback is welcome and aggregated in the open: open an issue or read the contributing guide.