MIT License Claude Code Just Markdown Obsidian Built from Daily Use

Claude Orchestrator
Starter Kit

Turn Claude Code into a personal AI assistant — with persistent memory, seamless session chaining, and structured workflows.

View on GitHub Quick Start ↓
You:

Without Orchestrator

  • 10 minutes re-explaining your project every morning
  • Context fills up — open a new terminal, start over
  • You build the same workflow for the third time
  • Quality varies — no feedback loop, just single-pass and hope
  • Knowledge scattered across tabs, notes, and memory

With Orchestrator

  • Sessions pick up where you left off
  • Three-layer memory persists context
  • Seamless session chaining when context fills up
  • Skills automate common workflows
  • Keyword routing invokes the right skill
  • Structured, consistent, quality-checked output

Not More Intelligence — Better Infrastructure

Claude is already smart. What it lacks is structure, memory, and habits.

Memory

Decisions, context, and preferences persist across sessions — not just within one.

Habits

Skills encode your best workflows. No more re-explaining the same steps every time.

Quality Loops

The Evaluator-Optimizer pattern catches issues that single-pass generation misses.

New

Session Chaining

Claude's context window is finite. Session Chaining makes it infinite — your state is saved automatically and the next session picks up where you left off.

55%
Context Monitor Statusbar tracks usage in real-time
Auto-Handoff Saves task, decisions & next steps
Smart Restart New session loads saved state automatically

Work without context limits

Use claude-loop as a drop-in replacement for claude. When context gets critical, your work is preserved and the next session continues seamlessly.

# Just use claude-loop instead of claude
claude-loop

# When context fills up:
# → State saved automatically
# → "New session with context? [Y/n]"
# → Press Y — picks up where you left off

How It Works

Your message flows through four stages — from intent to structured output.

Your Message "Summarize this article" CLAUDE.md — The Brain Routes keywords → skills, loads context Skill Execution Matched skill runs its workflow Three-Layer Memory Short · Mid · Long-term persistence Structured Output Result + next step suggestion feedback loop

Three-Layer Memory

Context that persists — from one conversation to months of knowledge.

LONG-TERM CLAUDE.md · memory/ · optional DB permanent MID-TERM Handoffs · Routing Log weeks – months SHORT-TERM Context Window
Short-term

Context Window

Current conversation and temporary results. Lifetime: 1 session.

Mid-term

State Files + Routing Log

Past session handoffs, routing decisions, and project states. Lifetime: weeks to months.

Long-term

CLAUDE.md + Memory Files + Knowledge Base

Routing rules, preferences, glossary, contacts, and domain knowledge. Lifetime: permanent.

Eight Core Skills

Reusable workflows defined as markdown files — instructions, not code.

capture

Quick Capture

Save notes and ideas quickly. Works with Obsidian, local files, or inline.

distill

Summarize & Condense

Progressive summarization. Extract key takeaways from any text.

express

Write Output

Generate polished, structured content — articles, posts, emails, concepts.

analyze

Deep Analysis

Structured ReAct analysis. Investigate topics with iterative reasoning.

signal-check

Quality Check

4-axis substance analysis: meaning, facts, framing, context. Signal vs. noise.

challenge

Stress-Test

Adversarial testing from 3 perspectives. Find weaknesses, blind spots, and hidden assumptions.

quality-gate

Quality Gate

Output quality orchestration with triage levels, external verification, and quality scoring.

handoff

Session Handoff

Save decisions, context, and next steps. Next session picks up seamlessly.

Four Key Feedback Loops

Skills chain together into powerful workflows.

Evaluator-Optimizer

Write, evaluate, improve — like a human review cycle.

express signal-check express evaluate optimize

Evaluator-Challenger

Write, attack, harden — adversarial stress-testing.

express challenge express stress-test harden

Evaluator-Gate-Optimizer

Write, verify with triage, revise — structured quality control.

express quality-gate express verify revise

Knowledge Cycle

Capture, process, output, verify — continuous learning.

capture distill express signal-check

This Is Just the Foundation

The starter kit gives you the building blocks. The full system goes much deeper.

YOU: "Analyze market data, write summary" CLAUDE.md — THE ORCHESTRATOR BRAIN Routing Rules User Patterns Workflow Chains matches "analyze" + "write" → chain: analyze → express SKILLS (20+ in full system, 8 in starter kit) capture distill express analyze signal-check quality-gate challenge handoff SHORT-TERM Context window 1 session MID-TERM Routing log · Handoffs weeks – months LONG-TERM CLAUDE.md · memory/ · Knowledge permanent Obsidian Your notes = Claude's KB SQLite DB Research · Datasets · Stats
Dimension Starter Kit Full System
Skills 7 core skills 20+ domain-specific skills
Routing 7 keyword mappings + mode detection 100+ mappings with disambiguation
Memory Template files Months of accumulated context
Obsidian 1 vault (optional) 6 connected vaults, 930+ notes
Database Empty schema Research items, datasets, usage stats
Workflow Chains 3 feedback loops 15+ multi-step chains
Self-Improvement Manual updates Automated routing analysis

Up and Running in 5 Minutes

# Clone the repo
git clone https://github.com/janrummel/claude-orchestrator-starter.git
cd claude-orchestrator-starter

# Copy to your Claude config
cp CLAUDE.md.example ~/.claude/CLAUDE.md
cp -r orchestrator/ ~/.claude/orchestrator/
cp -r memory/ ~/.claude/memory/
mkdir -p ~/.claude/orchestrator/projects

# Set up hooks (statusbar, context monitor, session start)
cp hooks/session-start.sh.example ~/.claude/hooks/session-start.sh
cp hooks/context-monitor.js ~/.claude/hooks/context-monitor.js
cp hooks/context-statusline.js ~/.claude/hooks/context-statusline.js
cp hooks/settings.json.example ~/.claude/settings.json
chmod +x ~/.claude/hooks/session-start.sh

# Optional: install session chaining wrapper
cp hooks/claude-loop.sh ~/.local/bin/claude-loop
chmod +x ~/.local/bin/claude-loop

# Start Claude Code
claude-loop
1 CLAUDE.md is read at startup — Claude understands its role
2 Hooks monitor context, auto-save state, and inject context at startup
3 Session chaining keeps you in the same terminal when context fills up

100% Local. Zero Telemetry.

Your Machine Only

All memory files, routing logs, and project states live in ~/.claude/. No server, no cloud upload, no analytics.

Plain Text, Full Control

Everything is markdown. Export, backup, version-control, or delete anytime. You own your data.

No Extra API Keys

The core kit needs zero API keys. Only Claude Code itself requires an Anthropic account.