CertPepper

Quick Start

# Clone and install
git clone https://github.com/cert-pepper/cert-pepper.git && cd cert-pepper
cp .env.example .env          # add ANTHROPIC_API_KEY for AI explanations
uv sync

# Set up a database and start studying
uv run cert-pepper db init
uv run cert-pepper ingest     # loads the bundled Security+ content
uv run cert-pepper study

Or open the repo in Claude Code and ask it to generate a question bank for any exam:

Set up a standard question bank for the CISSP exam

That request should map to the content MCP server's create_question_bank workflow. It creates or reuses a DB-backed question bank; it does not create a new examples/ content pack unless you explicitly ask for repository content files.

See MCP Integration to enable the servers.

Why CertPepper?

Most study tools — Anki, Quizlet, flashcard apps — treat every question the same. CertPepper adapts:

• FSRS-4.5 spaced repetition schedules reviews at optimal intervals based on your recall history — the same core algorithm behind Anki, tuned for exam prep.
• Bayesian Knowledge Tracing estimates mastery per domain and steers new questions toward your weakest areas, weighted by how much each domain counts on the real exam.
• AI explanations break down wrong answers with domain-specific context via Claude (MCP sampling in Claude Code, or the Anthropic API in CLI mode).
• Domain-weighted scoring predicts your exam score using actual exam weights, with a conservative prior for unseen material.
• MCP integration — study inside Claude Code with three servers for content, study sessions, and analytics.

Features

No guarantees. Predicted scores and pass probabilities are estimates based on practice performance — not predictions of real exam outcomes.

MCP Integration

Three STDIO MCP servers are registered in .mcp.json:

Enable them in Claude Code by adding enableAllProjectMcpServers: true to .claude/settings.local.json.

In Claude Code, explanations use MCP sampling — no ANTHROPIC_API_KEY needed.

For exam setup, the content server exposes two equivalent tools:

• create_question_bank(exam_name, size=None) — preferred user-facing entrypoint
• setup_exam(exam_name, size=None) — lower-level alias with the same behavior

Both tools accept size tiers lite, standard, and heavy. They create or reuse DB-backed question banks rather than writing content under examples/.

Adding Your Own Exam

1. Create a content directory in CertPepper's markdown format.
2. Set CONTENT_ROOT=/path/to/your/content in .env.
3. Run cert-pepper db init and cert-pepper ingest.

The examples/security-plus/ directory is a complete reference implementation with 228 questions, 135 flashcards, and 262 acronyms.

Configuration

Repository Layout

cert-pepper/
├── cert_pepper/          — Python source (CLI, MCP servers, algorithms)
│   ├── cli/              — Typer commands
│   ├── mcp/              — Three FastMCP STDIO servers
│   ├── engine/           — FSRS, BKT, selector, scorer (pure Python)
│   ├── ingestion/        — Markdown parsers
│   ├── ai/               — Anthropic client + explainer
│   └── db/               — SQLAlchemy async engine + schema
├── tests/                — pytest suite
├── examples/
│   └── security-plus/    — Security+ SY0-701 exam content
└── docs/
    ├── security-plus-in-10-days.md — 10-day Security+ study guide
    └── content-format.md — Format spec for your own exam content

Worked Example

The examples/security-plus/ directory contains a complete Security+ SY0-701 exam prep set:

• 5 domains of notes
• 135 flashcards
• 228 practice questions across all 5 domains
• 262 acronyms

See Security+ in 10 Days for a step-by-step guide.

Contributing

See CONTRIBUTING.md for setup instructions, workflow, and PR guidelines.

License

MIT