INSTRUCT
28%
Does the agent understand what we want?
0%
0 / 28 pts
Four axes
30%
Can the agent find its way around & use the right tools?
VALIDATE
30%
Can the agent tell if it did it right?
0%
0 / 30 pts
SECURE
12%
Can the agent run safely? (the 4th axis v2 added)
0%
0 / 12 pts
Overall score and radar chart
Overall score
0
Below L1
Toggle checks above to see your maturity level.
L1 40L2 55L3 70L4 85L5 95
Four-axis radar
liveRadar chart unavailable offline. Axis values:
The agent-ready-skill v2 rubric scores a codebase across 7 evidence-based dimensions, each weighted by how much it actually affects an AI agent's ability to work autonomously. Every sub-criterion is tagged portable (counts for any agent) or target (only scored when you declare an agent via --agents) — the dimmed checks below are target-specific.
For this workshop we remap those 7 dimensions onto 4 axes built around the question an agent silently asks: understand what to do (INSTRUCT, 28), find where to do it & reach for tools (NAVIGATE, 30), verify it did it right (VALIDATE, 30), and run safely (SECURE, 12). The model started as three axes — building v2, security forced a fourth. The 4 axes sum to 100.
Your overall score is the raw sum of the points you toggle on. Cross the thresholds 40 / 55 / 70 / 85 / 95 and you climb the maturity ladder L1 to L5.
| Axis | Dimension | Pts |
|---|---|---|
| INSTRUCT | D1 Agent Instructions & Context | 18 |
| D7 Spec-Driven Workflow & Docs | 10 | |
| NAVIGATE | D2 Navigability & Code Intelligence | 18 |
| D5 Agent Tooling & Capabilities | 12 | |
| VALIDATE | D3 Testing & Feedback | 16 |
| D4 CI/CD, Automation & Governance | 14 | |
| SECURE | D6 Security & Sandbox | 12 |