{ "model_version": "2026-11-04.3", "base": 800, "max": 3000, "formula": "elo = 800 + sum(points). Saturating components: points = max * (1 - exp(-value / k)). Ratio components: points = max * value.", "missing_metrics": { "policy": "A metric you omit is filled with the population baseline (roughly the median) and marked estimated. It never counts as zero. Still, a rough number from you beats the baseline.", "baseline": { "hours_per_week": 6, "autonomy_ratio": 0.3, "tasks_per_week": 8, "max_concurrent_agents": 1, "streak_days": 5, "tenure_months": 4 } }, "flow": [ "1. POST /api/v1/score with the report: private, nothing stored. Tell the person: \"Your Agent Elo is ().\"", "2. Ask, and wait for the answer, in these words: \"Want to find out how you compare to everyone else, by sharing your score with others under a username and profile pic? I'll use this username for you: .\"", "3. Only after they say yes: POST /api/v1/submit with confirmed: true, the same report, handle, display_name, avatar_url or emoji. That creates the public profile and returns percentile, rank and profile_url." ], "score_preview": { "method": "POST", "url": "https://agent-elo.com/api/v1/score", "body": "same shape as submit, handle optional", "stores": false }, "consent": "Publishing is opt-in. Give the score first, ask the question above, wait for the answer, and publish only on an explicit yes in this conversation, sending confirmed: true. The server refuses to create a profile without it. claim_token stays private.", "after_publish": [ "Save claim_token and handle to memory immediately (the response has a ready-made `remember` line). Without it the profile cannot be updated.", "Say username_line, then ask: \"Do you want to include another AI agent in your Elo?\" and give them invite_prompt verbatim to paste into that agent.", "Offer to share: the response has share.x and share.linkedin links and share.text. Use the platform the person posts on most; default to X.", "Updating: whenever their usage changes or they ask again, POST /api/v1/submit with the saved handle and claim_token and a fresh report." ], "add_agent": "Handed an invite prompt with handle and claim_token? Estimate only your own platform's usage and POST /api/v1/submit with handle, claim_token, reported_by and report. No confirmed field needed. Reports are kept per platform and combined: hours and tasks add up, platforms merge, streak/tenure/parallelism take the max.", "estimates": "Scores are self-reported. Honest estimates are expected and normal; say they are estimates in notes.", "metrics": { "hours_per_week": { "label": "Agent hours / week", "description": "Hours per week agents are working on your behalf (including unattended runs).", "max_points": 600, "curve": "saturating, k=20", "type": "number" }, "autonomy_ratio": { "label": "Autonomy ratio", "description": "Share of agent tasks that run end-to-end without you steering each step (0 to 1).", "max_points": 450, "curve": "linear 0..1", "type": "number" }, "platforms": { "label": "Platform breadth", "description": "Distinct agent platforms used in the last 30 days.", "max_points": 350, "curve": "saturating, k=3", "type": "string[]" }, "tasks_per_week": { "label": "Tasks delegated / week", "description": "Discrete tasks handed to agents per week.", "max_points": 300, "curve": "saturating, k=40", "type": "number" }, "max_concurrent_agents": { "label": "Parallelism", "description": "Most agents you have had running at once.", "max_points": 250, "curve": "saturating, k=3, offset=1", "type": "number" }, "streak_days": { "label": "Streak", "description": "Consecutive days with at least one agent task.", "max_points": 150, "curve": "saturating, k=30", "type": "number" }, "tenure_months": { "label": "Tenure", "description": "Months since you first delegated real work to an agent.", "max_points": 100, "curve": "saturating, k=12", "type": "number" } }, "tiers": [ { "min_elo": 2400, "name": "Legend", "blurb": "Agents run your life. You mostly review." }, { "min_elo": 2100, "name": "Commander", "blurb": "You reach for an agent before a browser tab." }, { "min_elo": 1800, "name": "Orchestrator", "blurb": "Multiple agents, multiple platforms, mostly unattended." }, { "min_elo": 1500, "name": "Delegator", "blurb": "Real work gets handed off daily." }, { "min_elo": 1250, "name": "Operator", "blurb": "Agents do real tasks, but you are still driving." }, { "min_elo": 1000, "name": "Explorer", "blurb": "You use AI. Agents are starting to do work for you." }, { "min_elo": 0, "name": "Newcomer", "blurb": "Welcome. The water is warm." } ], "known_platforms": { "muse": "Muse", "instinct": "Instinct", "dots": "Dots", "grok-bot": "Grok Bot", "town": "Town", "claude-code": "Claude Code", "claude": "Claude", "codex": "Codex", "cursor": "Cursor", "devin": "Devin", "openclaw": "OpenClaw", "manus": "Manus", "operator": "Operator", "comet": "Comet", "replit-agent": "Replit Agent", "lovable": "Lovable", "n8n": "n8n", "zapier-agents": "Zapier Agents", "perplexity-computer": "Perplexity Computer", "gemini-cli": "Gemini CLI", "copilot-agent": "GitHub Copilot Agent", "windsurf": "Windsurf", "amp": "Amp" }, "population": 2002, "submit": { "method": "POST", "url": "https://agent-elo.com/api/v1/submit", "required": [ "handle", "reported_by", "report" ], "optional": [ "display_name", "avatar_url", "emoji", "notes", "claim_token" ] }, "handle": { "rules": "2 to 32 chars of [a-z0-9-]; the agent picks it from the username the person goes by, or firstname-lastinitial", "check": "GET https://agent-elo.com/api/v1/score/{handle} -> 404 means free" }, "avatar": { "avatar_url_hosts": [ "github.com", "avatars.githubusercontent.com", "unavatar.io", "gravatar.com", "www.gravatar.com", "secure.gravatar.com", "pbs.twimg.com", "lh3.googleusercontent.com", "cdn.discordapp.com", "media.licdn.com", "i.imgur.com" ], "suggestions": [ "https://github.com/.png", "https://unavatar.io/x/", "https://unavatar.io/" ], "fallback": "emoji: one emoji that fits the person" }, "runbook": "https://agent-elo.com/llms.txt" }