AI Maturity ScannerLocal CLI for AI coding maturity
中文EN

Metrics

AMI is based on repository assets, not guesswork.

The scanner aggregates 15 raw metrics, normalizes them into three weighted dimensions, then produces a 0-100 AMI score and an L0-L4 maturity level.

Repository AI Maturity ai-maturity-scanner Algorithm v1 · Jul 15, 2026, 08:00 · abc12345 Profile Skill Workshop Structured Context·HighDeep Tool Connectivity·MediumCross-Project Coverage·Medium AMI score 67.5 AI Maturity L3 Proficient AI files 44 Ability applied 43 AMI score 67.5 Configuration depth 75.0 Context richness 60.0 Integration breadth 40.0 Skills 83 Advanced skills 30 Agents 30 Commands 10 MCP servers 60 Instructions 100 Spec files 100 Coverage 60 Scan my repository AI maturity Scan metric sources

Image score

How to read the image report scores

The image report shows normalized scores. Each metric is converted to a 0-100 score, and the progress bar visualizes that score. AI files and ability applied are raw counts, not progress-bar scores.

Skills

skill_count

A score (max 100) derived from the number of skills.

Advanced skills

advanced_skill_count

A score (max 100) derived from advanced skills.

Agents

agent_count

A score (max 100) derived from agent definitions.

Commands

command_count

A score (max 100) derived from command files.

MCP servers

mcp_count

A score (max 100) derived from unique MCP server names.

Instructions

ai_instruction_files

A score (max 100) derived from AI instruction files.

Coverage

subproject_coverage

A score (max 100) derived from covered subprojects.

Configuration depth

60%

The count and structure of skills, skill resources, agents, commands, and MCP files.

Context richness

30%

Stable model context from AI instruction files and specs.

Integration breadth

10%

Subproject-scoped instruction coverage across larger repositories.

The 15 metrics

skill_count

Number of skills in the repository.

skill_line_count

Total lines across skill files.

advanced_skill_count

Number of advanced skills.

skill_engineering_rate

Share of skills that are advanced.

skill_resource_count

Number of resource files under skills/.

agent_count

Number of agent definitions.

agent_line_count

Total lines across agent files.

command_count

Number of command files.

command_line_count

Total lines across command files.

mcp_count

Number of unique MCP server names in repository-level MCP config.

ai_instruction_files

Number of AI instruction files (e.g. CLAUDE.md, AGENTS.md).

instruction_max_line_count

Line count of the longest instruction file.

specs_line_count

Total lines across spec/document files.

subproject_coverage

Number of apps/ or libs/ subprojects with an AI instruction file.

L0-L4 level

L0
Not Started
No AI instruction file is present.
L1
Beginner
The baseline once the repository has basic AI collaboration signals.
L2
Improving
A repository with meaningful capability assets and at least one advanced skill.
L3
Proficient
A stronger repository with capability assets, advanced skills, specs, and engineering depth.
L4
Expert
A mature repository where AI capability assets and specs are present at scale.