C
Give Claude Code persistent memory across all sessions
Get Started
Claude Code Plugin

End Session Amnesia Forever

Claude Code is powerful, but every session starts fresh. Onelist gives it persistent memory that spans sessions, projects, and time.

The Problem: Every Session Starts Fresh

Without persistent memory, you repeat yourself constantly.

Standard Claude Code

# Session 1
You: "Use PostgreSQL, not MySQL, because of JSONB support"
Claude: "Got it, I'll use PostgreSQL for JSONB."
# Session 2 (next day)
You: "Let's continue with the database"
Claude: "What database are you using? MySQL or PostgreSQL?"
// Claude has no memory of yesterday's decision

Time Lost Per Session

  • 10 min Re-explaining project context
  • 15 min Finding past decisions
  • 30 min Re-debugging known issues

Hidden Costs

  • Inconsistent suggestions
  • Lost tribal knowledge
  • Repeated mistakes

The Solution: Persistent Memory

Onelist acts as Claude Code's external long-term memory.

With Onelist Integration

# Session 2 (next day)
You: "Let's continue with the database"
Claude: [context auto-injected from Onelist]
Claude: "I see from your memories that you chose PostgreSQL
for JSONB support. The schema is in db/schema.sql.
Last session you added the users table. Continue?"
// Claude remembers everything

How It Works

1. Session Start

When Claude Code launches, Onelist injects relevant memories scoped to your current project.

<onelist-context>
## Recent Memories
- PostgreSQL chosen for JSONB
- Auth uses JWT + Redis
</onelist-context>

2. During Session

Every file edit, command, and task is captured silently in the background. Search past memories anytime.

/onelist:search auth flow
Found 3 memories:
- JWT implementation (Jan 15)
- Redis session store (Jan 12)

3. Session End

When you stop, a summary is automatically saved with files modified, actions taken, and project context.

## Session Summary
- Edit: 12 files
- Added OAuth support
- Fixed login redirect bug

Standard vs. Onelist-Enhanced

Capability Standard Claude Code With Onelist
Session Memory Current session only Persistent forever
Project Context Manual explanation each time Auto-injected at start
Decision History Lost after session Searchable forever
Cross-Session Search Not possible Full-text + semantic
Session Summaries Copy/paste manually Automatic with metadata
Team Knowledge Not supported Shared memory pools
Codebase Familiarity Starts fresh each time Builds over time

Real-World Use Cases

🐛

Debugging Complex Issues

When a bug looks familiar, search your memories instead of re-investigating.

You: "This error looks familiar"
Claude: [searches Onelist]
"Found 3 sessions with this error.
Root cause was race condition in
scheduler.ex:142 (Feb 1 session)"
👋

Team Onboarding

New team members can search past sessions to understand architectural decisions.

New Dev: "Why JWT over sessions?"
Claude: [searches Onelist]
"Decision made Oct 15. Reasons:
1. Stateless for horizontal scaling
2. Mobile app compatibility"
🔄

Context Switching

Work on 5 projects without mixing up approaches. Memory is scoped per project.

# Project A context injected
- Uses Redis for caching
- REST API, no GraphQL
# Project B context injected
- Uses Memcached
- GraphQL only
📅

Long-Running Projects

After 6 months and hundreds of sessions, Claude still knows the full history.

# 6 months of accumulated memory
- 847 sessions captured
- 2,341 files modified
- 156 architectural decisions
- All searchable instantly

Quantified Benefits

For a developer with 5 Claude Code sessions per day

50 min
saved daily
4 hours
saved weekly
16 hours
saved monthly

Where the time goes

Context setup
10 min0
Finding decisions
15 min2
Re-debugging
30 min5

Your Data, Your Control

Self-host everything. Nothing leaves your machine unless you want it to.

Self-Hosted Option

Run Onelist-Local entirely on your infrastructure. Docker or native. Your database, your rules.

End-to-End Encryption

Data encrypted at rest and in transit. API key authentication. Full deletion capability.

No Model Training

Your memories never train any models. Your code stays your code.

Open Source

Audit the code yourself. MIT licensed. Community-driven development.

Get Started in 5 Minutes

1

Install the Plugin

# Link to Claude Code plugins
ln -s /path/to/onelist-local/extensions/claude-code ~/.claude/plugins/onelist
2

Start Onelist-Local

docker-compose up -d
3

Connect

# In Claude Code
/onelist:connect
URL: http://localhost:4000
Key: your-api-key

Start Coding

Memories auto-inject. Sessions auto-save. Search with /onelist:search.

Stop Re-Explaining. Start Building.

Give Claude Code the memory it deserves. Free with Onelist-Local.