Give your AI a memory. Works with Claude Desktop and any MCP-compatible client.
Add to your Claude Desktop config:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"contextable": {
"command": "npx",
"args": ["@contextable/mcp"]
}
}
}Restart Claude Desktop. Done.
Now ask Claude: "Create a project called My App to track my development decisions"
Save context that persists across conversations:
- "Save this architecture decision to the project"
- "Remember this API design for later"
- "Store this bug investigation so I don't forget"
Find anything instantly:
- "Search for authentication"
- "What did we decide about the database?"
- "Load all security-related artifacts"
Pick up where you left off:
- "Resume my project"
- "Show me the project summary"
- "What were we working on?"
| Feature | Description |
|---|---|
| Projects | Organize context by project or topic |
| Artifacts | Save decisions, code snippets, docs, conversations |
| Full-Text Search | Find anything across all projects |
| Version History | Every change tracked, rollback anytime |
| Auto-Chunking | Large content automatically split |
| Topic Clustering | Auto-detect themes across artifacts |
project_save- Create or update a projectproject_list- List all projectsproject_resume- Load project with summariesproject_analysis_get- Get AI-generated insights
artifact_save- Save content (auto-chunks large files)artifact_list- List with size estimatesartifact_get- Load full contentartifact_delete- Archive (recoverable)artifact_restore- Restore archivedartifact_archived- List archived itemsartifact_versions- View historyartifact_rollback- Restore previous version
search- Full-text search across everything
Everything is stored locally in SQLite:
~/.contextable/data.db
No account. No cloud. No tracking. Your data stays on your machine.
Contextable Cloud adds:
| Feature | Local | Cloud |
|---|---|---|
| Claude Desktop | ✓ | ✓ |
| ChatGPT | - | ✓ |
| Claude.ai (web) | - | ✓ |
| Sync across devices | - | ✓ |
| AI analysis & insights | - | ✓ |
| Team sharing | - | Coming soon |
| Variable | Default | Description |
|---|---|---|
CONTEXTABLE_DATA_DIR |
~/.contextable |
Data directory |
CONTEXTABLE_DB_PATH |
~/.contextable/data.db |
SQLite database path |
CONTEXTABLE_LOG_LEVEL |
info |
Logging: debug, info, warn, error |
{
"mcpServers": {
"contextable": {
"command": "npx",
"args": ["@contextable/mcp"],
"env": {
"CONTEXTABLE_DB_PATH": "/path/to/my/data.db"
}
}
}
}For web-based MCP clients, run as HTTP server:
npx @contextable/mcp --sse --port 3000Endpoints:
GET /health- Health checkPOST /mcp- MCP messages (JSON-RPC)
import { SQLiteAdapter } from '@contextable/mcp';
const storage = new SQLiteAdapter({ path: './my-context.db' });
await storage.initialize();
// Create a project
const project = await storage.createProject({
name: 'My Project',
description: 'Project description',
});
// Save an artifact
const artifact = await storage.createArtifact({
projectId: project.id,
title: 'Design Decision',
artifactType: 'decision',
content: '# We chose PostgreSQL because...',
});
// Search
const results = await storage.search('postgresql');
await storage.close();git clone https://github.com/Contextable-me/mcp.git
cd mcp
npm install
npm run build
npm testApache 2.0 - See LICENSE
- Contextable Cloud - Sync, ChatGPT, AI analysis
- GitHub - Source code
- MCP Protocol - Specification
- npm - Package