brain-jar
Claude Code plugins that remember, search, investigate, visualize, and discover.
Claude is brilliant but forgetful. Every session starts from zero—no memory of your preferences, no context from yesterday, no accumulated knowledge. Brain-jar fixes that.
The Plugins
| Plugin | What It Does |
|---|
| shared-memory | Persistent memory and user profiles across sessions |
| perplexity-search | Web search that knows your stack and history |
| forensics | Guided reverse engineering for APIs, protocols, and code |
| visual-thinking | Persistent diagrams that Claude can read, build on, and export |
| pattern-radar | Personalized trend detection at the intersection of your knowledge |
Why Brain-Jar?
Before
Claude: What framework are you using?
You: React with TypeScript (for the 50th time)
Claude: Here's how to do authentication—
You: I already solved this last week, different project
Claude: Let me search for that...
[Returns Java tutorials when you use Node]
After
Claude: I see you're using React/TypeScript. Based on your
auth work last week, here's a pattern that fits...
[Searches return Node-specific results]
Claude: Your investigation from yesterday found 4 endpoints.
Ready to build the OpenAPI spec?
Installation
Add brain-jar as a Claude Code marketplace:
/plugin
→ Add Marketplace
→ designnotdrum/brain-jar
Install plugins individually:
/plugin
→ brain-jar
→ [Select plugin]
The Stack
┌─────────────────────────────────────────────────────────────────────────┐
│ Claude Code │
├─────────────────────────────────────────────────────────────────────────┤
│ shared-memory perplexity-search forensics │
│ ─────────────── ───────────────── ────────── │
│ • User profiles • Context-aware • Investigations │
│ • Memory store • Search history • Traffic analysis │
│ • Auto-summaries • Profile-enriched • Spec generation │
├─────────────────────────────────────────────────────────────────────────┤
│ visual-thinking pattern-radar │
│ ─────────────── ───────────── │
│ • Mermaid diagrams • HN/GitHub scanning │
│ • Version history • Pattern detection │
│ • draw.io export • Profile-aware relevance │
└─────────────────────────────────────────────────────────────────────────┘
↓
┌────────────────────────┐
│ @brain-jar/core │
│ Shared Mem0 client │
│ Common types │
└────────────────────────┘
↓
┌────────────────────────┐
│ Mem0 Cloud │
│ Persistent storage │
└────────────────────────┘
All plugins share the same user profile and memory infrastructure. Install shared-memory first for the best experience.
Highlights
shared-memory v2.2.2
- First-run prompt - New users are automatically prompted to enable chess timer hooks on first launch (opt-in)
- Auto-install hooks - Chess timer bundles hookify rules via setup_chess_timer_hooks tool for fully automatic session management
- Chess timer - Self-calibrating time estimation that tracks how long you actually spend coding, predicts future feature duration based on complexity and work type
- 7 new MCP tools - start_work_session, pause/resume/complete, get_work_estimate, list_work_sessions, get_active_session
- Confidence levels - Predictions improve with more samples (low <5, medium 5-14, high 15+)
- 24 MCP tools for memory, profiles, summaries, diagnostics, and work sessions
perplexity-search v2.0.2
- Subagent pattern with TL;DR - skill shows condensed summary + full Perplexity response
- Profile-enriched queries return stack-relevant results
- Search memory builds context over time
- Perplexity Sonar for real-time, cited answers
- Bundled dependencies for instant marketplace install
