Free & Open Source

Audit Your Agentic Skills

Run a deep audit of your AI engineering setup. One command. Instant score. Actionable recommendations.

Install the Skill

Add the audit skill to any project that uses Claude Code. It runs inside your IDE or terminal and scans your agentic configuration.

Step 1: Clone the skill
$git clone https://github.com/oaseru/agentic-skill-auditor.git
Step 2: Copy the skill to your project
$cp -r agentic-skill-auditor/.claude/skills/audit-skill .claude/skills/
Step 3: Run the audit in Claude Code
$/audit-skill

Quick start (no install)

$npx @oaseru/agentic-audit

What It Scans

The auditor evaluates 9 categories of your agentic engineering maturity, weighted by impact.

1.5x

Context Setup

CLAUDE.md files, project instructions, and context configuration

1.0x

Tool Connections / MCP

MCP server configs, connected tools, and external integrations

1.0x

Skills & Commands

Custom slash commands, skill definitions, and reusable prompts

1.0x

Memory & Compounding

Persistent memory, knowledge banks, and cross-session learning

1.5x

Feedback Loops & Hooks

Pre/post hooks, quality gates, and automated feedback systems

1.0x

Pipeline / Headless

CI/CD integration, headless execution, and batch processing

1.0x

Multi-Agent / Multi-Model

Agent orchestration, model routing, and parallel execution

1.0x

Execution Quality

Error handling, rollback strategies, and output validation

2.0x

Autonomous Workflows

End-to-end task completion without human intervention

How It Works

01

Scan Your Project

The skill reads your .claude/ directory, CLAUDE.md files, MCP configs, hooks, skills, and memory files to understand your agentic setup.

02

Score Across 9 Categories

Each category is scored and weighted. Your total maps to a level on the 0–900 Agentic Engineering Scale.

03

Get Recommendations

Receive specific, actionable steps to level up. The audit tells you exactly what to add or configure next.

Example Output

# Agentic Engineering Audit Report
Overall Score: 642 / 900
Level: 7 — Orchestrator
Category Breakdown:
Context Setup .............. 92/100 (1.5x)
Tool Connections / MCP ..... 85/100
Skills & Commands .......... 78/100
Memory & Compounding ....... 65/100
Feedback Loops & Hooks ..... 60/100 (1.5x)
Pipeline / Headless ........ 55/100
Multi-Agent / Multi-Model .. 72/100
Execution Quality .......... 80/100
Autonomous Workflows ....... 40/100 (2.0x)
Next Steps:
1. Add pre-commit hooks for automated quality gates
2. Configure headless mode for CI/CD pipeline runs
3. Set up autonomous task completion with approval flow

Take the Web Audit

No install required. 10 questions, 2 minutes.

Audit Me