IDE Plugins

VS Code Extension (v0.9.7)#

Features

• Status Bar — live lesson count and brain health
• Health Panel — overview with token savings and cost estimates
• Lesson Viewer — every lesson with recall counts and what worked
• Token Savings — roughly 1,200 tokens saved per recall
• Auto-refresh — configurable interval (default: 5 minutes)
• Offline Queue — lessons saved locally when the Brain is unreachable, auto-synced on reconnect

Download: cachly-brain-0.9.7.vsix

Setup

1. Download the .vsix file
2. In VS Code: Cmd+Shift+P → "Extensions: Install from VSIX…"
3. Open Settings → search "Cachly"
4. Set your API Key (cky_live_...)
5. Set your Instance ID (UUID from dashboard)

Copy for your AI — paste this into Copilot Chat, Claude, Cursor, or Windsurf and your AI sets everything up:

Install the Cachly Brain VS Code extension, then configure it:
1. Download and install the VSIX from https://cachly.dev/downloads/cachly-brain-0.9.7.vsix
2. Open VS Code Settings (Cmd+, or Ctrl+,) and search for "Cachly"
3. Set cachly.apiKey to your API key (starts with cky_live_...)
4. Set cachly.instanceId to your Brain instance UUID (from the Cachly dashboard)
5. The status bar should now show "Brain: N lessons"
6. Use Cmd+Shift+P → "Cachly: Show Brain Health" to see full stats
7. Use Cmd+Shift+P → "Cachly: Show Lessons" to see all learned lessons

Configuration

IntelliJ / JetBrains Plugin (v0.3.0)#

Install from the JetBrains Marketplace.

Features

• Status Bar Widget — live lesson count
• Health Dialog — lesson table with recall counts and token savings
• Brain Doctor — one-click diagnostics for API key, instance ID, and connectivity
• Settings Panel — Tools → Cachly Brain

Setup

1. Settings → Plugins → Marketplace → search "Cachly Brain" (or install from the web)
2. Go to Settings → Tools → Cachly Brain
3. Enter your API Key and Instance ID
4. Run Tools → Run Brain Doctor to verify the connection

Build from source

cd sdk/intellij
./gradlew buildPlugin
# Output: build/distributions/cachly-brain-intellij-0.3.0.zip

How lessons work#

Lessons are created when an AI assistant — Copilot, Claude, Cursor, Windsurf — calls learn_from_attempts() through the Cachly MCP server.

1. AI learns. After fixing a bug, your AI calls learn_from_attempts() to store what worked.
2. AI recalls. Next session, recall_best_solution() returns the fix instantly, no re-research.
3. You save. Roughly 1,200 tokens per recall. Over time, that adds up to thousands of dollars and hours saved.

Related#

Both plugins read from the same Brain instance you set up in 3-Layer AI Memory System — no separate configuration. See MCP Integration for how the underlying session lifecycle feeds the lesson counts shown in each status bar.

← Back to all docs