Measure — Agentic Level

How agentic is your life? Measure it.

Your AI agent reads one instruction file, scans your setup, and writes a profile — counts and hashes, never secrets. This page turns it into a score in person-equivalents — how many people's worth of work your system does — and a band on the cosmic scale, from Asteroid to Supercluster.

  1. 1. Get SOFTCYB.mdOne instruction file. It tells your agent what to look at and how to write it down — counts and hashes, never secrets.
  2. 2. Feed it to your agentPaste it into Claude Code, Codex, Cursor, OpenClaw — whatever runs your life. It scans your setup and writes softcyb-profile.md.
  3. 3. Upload the profileDrop the file on /level. Your score in person-equivalents, your band on the cosmic scale, and where you stand.
Get SOFTCYB.mdUpload profile
Your profile file

Drop softcyb-profile.md here

Read locally. Nothing is sent anywhere until you press Save or Share.

How the score works (provisional rubric)
Dimension0 means5 means
Automation depthnothing runs unless you typeroutines watch routines; the schedule edits itself
Tool connection breadthmodel onlyeleven or more, and the agent has added one itself
Agent autonomychat only; you copy results by handa 24/7 brain merges or deploys under a written policy
Memory and long contextevery conversation starts from zeroshared, indexed memory across agents and devices
Embodimentkeyboard and screenambient: glasses, ring, watch, replies on the body
Self-modificationyou edit all configurationit maintains itself under a written policy
Life penetrationnone beyond curiosityeight or more; the agent is the first stop for anything

Bands · person-equivalents (how many people's worth of work)

  1. Level 1 — AsteroidL1 Asteroid< 0.1
  2. Level 2 — MoonL2 Moon0.1 – 0.3
  3. Level 3 — PlanetL3 Planet0.3 – 0.8
  4. Level 4 — StarL4 Star0.8 – 3
  5. Level 5 — Solar SystemL5 Solar System3 – 8
  6. Level 6 — GalaxyL6 Galaxy8 – 20
  7. Level 7 — SuperclusterL7 Supercluster20 +

Gates: galaxy needs no dimension at 0; supercluster needs every dimension at 4 or above.

The seven dimensions (0–5 each, 35 points) are the grounds; the score is the estimate they support: person-equivalents, read as role coverage × parallel workers. Every score of 3 or more must cite evidence; declared points count half and cap a role at “on request”. The rules follow docs/spec/AGENTIC-LEVEL.md v0 with the person-equivalent bands of AGENTIC-LEVEL-v1-PLAN.md §5.0, and live in one file (src/data/rubric.ts).

Everything runs in your browser. The file never leaves your machine.rubric 0.2