How Claude Certification Tracks Are Structured Using tracks/*.json

Claude certification tracks in the rohitg00/ai-engineering-from-scratch repository are defined by JSON manifest files located in certifications/claude/tracks/ that encode exam metadata, weighted knowledge domains, sequential lesson paths, and personalized study schedules in a machine-readable schema.

The rohitg00/ai-engineering-from-scratch repository implements a data-driven approach to curriculum management through structured JSON manifests. Instead of hard-coding certification logic, the system uses discrete track definition files to render study roadmaps, generate exam items, and drive the learner interface dynamically. Understanding the schema behind these tracks/*.json files is essential for developers extending the platform or integrating Claude certification data into custom tooling.

Anatomy of a Track Manifest File

Each JSON file in certifications/claude/tracks/ (such as ccar-f.json) follows a uniform schema containing nine top-level sections that completely define a certification track.

Track Identification and Metadata

The manifest begins with core identifiers: id (unique internal identifier like "claude-ccar-f"), slug (URL-friendly name), examCode (official designation such as "CCAR-F"), and credential (human-readable title like "Claude Certified Architect – Foundations"). The summary field provides a one-sentence description of the track's focus area, such as making production tradeoffs across Claude Code, the Agent SDK, API, and MCP.

Knowledge Domains and Weighting

The domains array lists core knowledge areas with assigned weights that determine exam composition. For example, "agentic-architecture-orchestration" might carry a weight of 27, indicating its proportional representation in the certification exam. This weighting system enables automated exam item generation based on domain importance.

Curriculum Structure via Lessons

The lessons array defines the ordered curriculum through objects containing:

• path: Directory location (e.g., "certifications/claude/lessons/00-certification-strategy")
• domains: Array of knowledge domains covered
• role: Classification as orientation, core, review, or capstone
• required: Boolean indicating mandatory completion

Supplementary Content and Assessments

Optional deepDives reference advanced lessons from the main curriculum, such as "phases/14-agent-engineering/28-orchestration-patterns", while the assessments array configures diagnostic and mock exams with paths to JSON exam definitions like certifications/claude/assessments/ccar-f/diagnostic.json.

Study Schedule Configuration

The studyPlans array offers structured preparation timelines, specifying durationDays, hoursPerWeek, and descriptive labels like "28-day scenario plan" to help learners pace their preparation.

Consuming Track Data Programmatically

The self-contained JSON structure enables downstream tools to ingest certification data without hard-coded logic.

Python – Load Track Metadata and Filter Core Lessons

import json
from pathlib import Path

track_path = Path("certifications/claude/tracks/ccar-f.json").read_text()
track = json.loads(track_path)

print(f"Track: {track['credential']} ({track['examCode']})")
print("\nCore Lessons:")
for lesson in track["lessons"]:
    if lesson["role"] == "core":
        print(f"- {lesson['path']}")

JavaScript – Extract Study Plans

import { readFileSync } from "fs";

const track = JSON.parse(
  readFileSync("certifications/claude/tracks/ccar-f.json", "utf-8")
);

console.log(`Study plans for ${track.credential}:`);
track.studyPlans.forEach(sp => {
  console.log(`• ${sp.label}: ${sp.durationDays} days, ${sp.hoursPerWeek}h/week`);
});

Bash – Quick Overview with jq

jq '.credential, .domains[].name, .lessons[].path' certifications/claude/tracks/ccar-f.json

Multi-Track Schema Consistency

All tracks adhere to the identical schema, differing only in field values. The repository includes:

• ccar-f.json: Foundations track manifest
• ccar-p.json: Professional track manifest
• ccdv-f.json: Developer-focused track manifest
• ccao-f.json: Architecture-oriented track manifest

This uniformity allows the certification engine to process any track generically, while global metadata in certifications/claude/program.json (containing badge URLs and overall descriptions) and prerequisite mappings in certifications/claude/prerequisites.json provide program-wide context.

Summary

• Claude certification tracks use JSON manifest files in certifications/claude/tracks/ to define complete certification metadata including exam codes, credentials, and summaries
• Each manifest includes weighted domains, sequenced lessons with pedagogical roles (orientation/core/review/capstone), optional deepDives, assessments, and studyPlans
• The uniform schema across all tracks (ccar-f.json, ccar-p.json, etc.) enables the certification engine to programmatically render study roadmaps and generate weighted exam items
• Machine-readable structure supports integration with CLI tools, CI validation scripts, and custom learning management systems without modifying source code

Frequently Asked Questions

What is the exact file path for Claude certification track definitions?

Track manifests are located in certifications/claude/tracks/ within the rohitg00/ai-engineering-from-scratch repository, with individual files named according to their exam codes (e.g., certifications/claude/tracks/ccar-f.json for the Architect Foundations exam).

How are lesson dependencies and ordering controlled in the track manifests?

The lessons array in each JSON file defines explicit sequencing through array order, with each lesson object specifying its path, covered domains, pedagogical role (orientation/core/review/capstone), and whether it is required for certification completion.

Can third-party tools consume these track manifests to build custom study planners?

Yes, the JSON schema is self-contained and documented through the existing track files, allowing external Python, JavaScript, or Bash scripts to parse domains, study plans, and lesson paths without modifying the source repository.

Where are the actual exam questions stored if tracks only reference assessments?

While track manifests in tracks/*.json reference assessments via the assessments array (pointing to files like certifications/claude/assessments/ccar-f/diagnostic.json), the specific exam item pools are stored separately in the referenced assessment JSON files, not in the track manifests themselves.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}