$ claude --continue
✓ CLAUDE.md loaded
✓ .claude/rules/ — 9 files
✓ .claude/reference/ — 4 files
standing instructions active.
claude knows the project now.
# project orchestrator
@import .claude/rules/
@import .claude/reference/
lean root. everything composes from here.
the operating layer multi-project operators rebuild from scratch on every project,
packaged once. drop it in. no rebuild.
New project, same setup. CLAUDE.md, the rules folder, the ticket system, built from scratch again.
Every project gets a slightly different version. Nothing carries forward. Same hours, same work, every time.
The kit is that setup, done once. Drop it into any project and the operating layer is already there. It holds up across model releases. Not tutorials.

// THE TWO MODES
MODE 01
Layered .claude/rules/ and .claude/reference/ structure with conditional @import composition. Turns Claude Code from smart autocomplete into a junior engineer with standing instructions. For any project where Claude is writing or modifying code.
MODE 02
Editorial discipline, framing locks, and read-before-write rules. For writing-heavy projects where Claude generates long-form content and consistency matters more than fluency.
MODE 03 · SHIPS WITH V2
Lessons-format CLAUDE.md, verification protocols, and the tool fallback ladder. For workflows where Claude orchestrates multi-step processes. Ships with v2 at no additional cost to v1 buyers.
free for v1 buyers
// THE UNIVERSAL LAYER
These ship as a separate layer because they apply regardless of mode.
u1
How Claude inventories a project before suggesting changes.
u2
Five-part structure for getting Claude to fix something specific without breaking adjacent things.
u3
The failure modes already hit so you don't have to.
u4
The hierarchy that keeps long projects from collapsing into noise.
// WHAT'S IN THE BUNDLE
unzip solo-operator-kit-v1.zip
✓ done
The starter CLAUDE.md the kit is built around. Yours free — grab it and drop it into any project.
Instant download, no spam. Privacy.
FOR
+ Multi-project operators running more things than one person should. Building the next project before finishing the last one.
+ Notion refugees, Obsidian refugees. Already running file-based workflows. Want their AI system to travel across projects, not live in another login.
+ Solo operators who want the operating discipline without rebuilding the infrastructure from scratch on every project.
NOT FOR
– Anyone who hasn't actually used Claude Code yet. Start there first.
– Large engineering orgs with established AI tooling. Your needs are different.
– Anyone looking for prompt packs. This is operating infrastructure.


Lucky. Solo dev.
Open the files, run them, and see if they hold up for your project. Any questions? Shoot me an email.
No. A tutorial teaches you Claude Code. This assumes you already run it. The kit is the files Claude reads every session: the rules folder, the reference layer, the @import composition. Not steps you follow once and close.
The patterns sit above any single model. They're about how you structure standing instructions, not which model reads them. When the model changes, the structure holds. And v2 ships free to v1 buyers, so updates reach you regardless.
No. Code mode and content vault mode are separate layers. Start with the one that matches what you're building, add the other when a project calls for it. The universal layer applies either way.
Most of them are prompt packs or setup tutorials. This is the operating layer underneath: the rules, reference, and @import composition that hold across projects and across model releases. You install it once and drop it into the next project. Nothing to relearn.
30 days, no questions asked, through your original Gumroad purchase.