agentry

Who lets their agent run?

Scores how autonomously you work — derived from real Claude Code and Codex session data, not self-reported. Run the prompt below in your agent to get scored. Submitting to the leaderboard is optional.

#HandleScoreProfileClientSubmitted
1simonkillie
45
Autonomous Operator
Claude Code13 Jun, 20:21 CEST
2radicalb0re
42
Copilot Collaborator
Claude Code16 Jun, 15:33 CEST
Get your score

Paste this into your agent to get your score — it will ask before uploading anything.

Run `npx agentry-cli scan` to measure my developer autonomy score from my local Claude Code/Codex logs, then show me the score breakdown. Do NOT submit yet. After showing it, ask me whether I want to publish it to the agentry leaderboard — only if I say yes, run `npx agentry-cli scan --submit`. Only numeric session metadata is ever transmitted; no prompts, code, or file paths.

Reads only event types, timestamps, and tool-call counts from ~/.claude/ and ~/.codex/ — no prompt text, code, or file paths ever leave your machine. Source code on GitHub.

Profiles
Fleet Orchestrator85–100Manages agents as persistent autonomous resources
Hands-Off Architect70–84Strategic planner steering agent swarms
Autonomous Operator45–69Delegates feature ownership with periodic oversight
Copilot Collaborator20–44Iterative human-agent dialogue on implementation
Hand-Coder0–19Agents assist with routine tasks only