Cursor Supermemory
https://github.com/supermemoryai/cursor-supermemory- Category
- AI Agents
- Rank
- No. 1499Tools index
Previous survey · No. 1512 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- supermemoryai
- GitHub
- 23 stars
- Date
About
Supermemory plugin for Cursor that gives the editor persistent long-term memory across sessions.
What it does
Cursor Supermemory connects Cursor sessions to a Supermemory account. At session start, it loads profile facts and project knowledge into the AI context. At session end, it can save qualifying conversations. MCP tools also let the AI search, add, list, inspect, configure, and forget memories directly.
Why it's ranked here
The tool covers both automatic recall and explicit memory management, which makes it useful beyond passive transcript storage. Repository-derived container identities reduce collisions between projects, while compatibility reads preserve older memories. Its value still depends on trusting an authenticated external service with selected coding context.
What's good
Personal and project memories share a repository container but remain separated through scope metadata. Repository identity uses a normalized Git remote and a stable hash, with local paths as fallback. Searches merge results across current and legacy containers, remove duplicates, and tolerate individual container failures when another succeeds.
Tradeoffs
Session capture sends conversation text to Supermemory and can include up to 100,000 characters. Automatic capture skips abnormal endings, missing transcripts, and conversations with fewer than two user turns. Authentication needs a browser callback and stores an API key locally. The supplied tests cover repository identity helpers, not the broader integration.
How to use it well
It best suits developers who repeatedly return to long-running repositories and want Cursor to retain decisions, preferences, and project facts. Use explicit project memories for durable team knowledge and personal scope for working preferences. It does not replace source control, issue tracking, or maintained technical documentation.
Technical notes+
package.json defines an ESM TypeScript package with Bun build and test scripts, an npm command-line entry, the MCP SDK, and the Supermemory SDK. src/cli.ts exposes MCP startup plus login, logout, and status commands. src/auth.ts runs a loopback callback on port 19878 and writes credentials with restrictive directory and file modes. src/hooks/session-start.ts loads profile and project data concurrently, while src/hooks/session-end.ts parses JSON or JSONL transcripts and writes deterministic session captures. src/client.ts performs scoped hybrid search, compatibility reads, result merging, deduplication, and partial-failure handling. src/tags.ts derives canonical repository tags from normalized remotes or resolved paths. src/tags.test.ts tests remote normalization, name sanitization, and hash length.
Observed
- License
- MIT
- Primary language
- TypeScript
- Packaging
- ES module npm package built with Bun
- Install surface
- npm package installation, with bunx available for authentication and MCP startup
- Interfaces
- Cursor session hooks, stdio MCP server, and command-line authentication commands
- MCP surface
- Tools for configuration, container inspection, search, add, list, forget, and profile retrieval
- Tests
- A Bun test file covers repository identity helpers
Read from README.md, package.json, src/cli.ts, src/auth.ts, src/tags.ts, src/client.ts, src/config.ts, src/context.ts, src/tags.test.ts, src/mcp-server.ts, src/hooks/session-end.ts, src/hooks/session-start.ts, bun.lock, .mcp.json, CLAUDE.md.
What it can do
Store code context across editor sessions
Code files and project context → Persistent memory storage
Retrieve previous session information
Current editing session → Historical context and code patterns
Maintain long-term project memory
Project files and development history → Accumulated knowledge base
Provide contextual code suggestions based on history
Current code being written → Context-aware code recommendations
Track coding patterns over time
User coding behavior and preferences → Pattern analysis and insights
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.