A Course and an Agent Store for Claude Code
Build your own AI assistant, and understand every piece.
A hands-on course and a library of ready-made agents, skills and commands for Claude Code. You build it on your own machine, test each part, and keep what you build.
The Problem
Sound familiar?
Lost Context
A new Claude Code session does not remember what you built in the last one, so you explain it again.
Copy-Paste Prompts
Your setup is a pile of prompts you paste in by hand. When one breaks, you do not know which piece to fix.
No Method
You get results, but you cannot repeat them. Each project starts from zero because nothing you learned is written down.
Hard To Trust
You cannot maintain or trust what you did not build. When it stops working, you are stuck.
What You Build
Four parts you build and understand.
Each part works on its own. Together they cover the whole workflow.
Memory
A searchable store of your decisions, fixes and patterns, so a new session can look them up. The Course shows you how to set it up on your own machine.
Helper Agents
Small assistants with one job each, such as writing code or reviewing a plan, and a lead agent that hands work to them. You get ready-made examples and learn how to write your own.
Skills
Reusable instructions that tell Claude Code how to do a task your way. The Agent Store has ready-made ones, and the Course shows how to turn your own repeat work into a skill.
Checks Before You Ship
A planning step before big changes and a review step after them, so you catch mistakes early. You learn how to add both.
What Is Inside
What you get.
- 57ready-made agents, skills and commands
- 8course modules
As of 2026-10-22: 14,000+ memories across dozens of projects.
Curriculum
8 modules.
From your first memory store to running your setup in production. Each module explains one piece, then you build it and test it.
01 Foundation: Memory System FREE
Set up a searchable memory store for Claude Code. Install the memory server, learn the save and look-up patterns, and set the routine each session follows.
02 Memory Architecture
Design your knowledge graph. Memory types, cross-project search, skill pattern extraction, deduplication strategies.
03 Context Loading
Design the boot sequence that makes every session start informed. CLAUDE.md as an operating system spec, routing tables, file-marker skill detection.
04 Delegation Architecture
Design the execution layer where Opus thinks and Sonnet executes. Five agent levels, stateless parallel execution, delegation specs precise enough to hand off.
05 Quality Gates
Mandatory planning gates. Pre-implementation audit chains. Plan-enforcer for drift detection. Checkpoint-based phase management.
06 Content Pipeline
Automate knowledge compounding. Pattern storage across sessions, a 3-observation threshold that separates signal from noise, automatic SKILL.md generation from matured patterns.
07 Cross-Project Intelligence at Scale
Work across many projects without losing context. Cross-project look-up, agents you reuse everywhere, and what changes as the number of projects grows.
08 Production Operations
Deploy, monitor, maintain. Docker orchestration, VPS deployment, security audits, compliance tracking.
Pricing
Choose your plan.
The Course is a one-time purchase at the price shown on the card. The Agent Store is monthly and you can cancel anytime. Both need Claude Code.
Course
One-time payment, 30-day money-back guarantee
Eight modules on running Claude Code as a system: persistent memory, delegation, quality gates and production operations.
- ✓ 8 structured modules with exercises
- ✓ CLAUDE.md and configuration templates
- ✓ A starter set of agents and skills
- ✓ 30-day money-back guarantee
Agent Store
Ready-made agents, skills and commands for Claude Code.
- ✓ 57 ready-made agents, skills and commands
- ✓ Plain files you can open, read and change
- ✓ Delivered with the accruedev command line tool
- ✓ Updates arrive when you run accruedev pull
FAQ
Common questions.
What exactly do I get?
Course buyers get 8 modules with exercises, templates and a starter set of agents and skills. Agent Store subscribers get 57 ready-made agents, skills and commands. Both are delivered into Claude Code with the accruedev command line tool.
Do I need Claude Code?
Yes. Everything here is built for Claude Code, Anthropic's command line coding assistant. You need your own Claude Code access (see Anthropic's site for which plans include it) and Node.js 20 or newer to install our tool.
Is this just a collection of prompts?
No. The Course explains how the pieces fit together (memory, loading context, handing work to helper agents, checks before you ship) and has you build each one. The Agent Store gives you ready-made agents, skills and commands you can open, read and change.
Will I understand what I install?
That is the point of the Course. Each module explains one piece, then you build it and test it yourself. Agents, skills and commands are plain text files in your ~/.claude folder, so you can open and read them.
Can I get a refund?
Yes. 30-day money-back guarantee on the Course, no questions asked.
What if Claude Code changes its file structure?
Files are written to ~/.claude/ and ~/accruedev/. If Claude Code changes its layout, some content may need an update. Run `accruedev pull` to get the latest versions.
Self-build kits for specific jobs, such as turning an inbox into approved tasks, are in preparation.