Skip to main content
Glama

claude-recall

Persistent memory plugin for Claude Code. Captures session context and injects relevant history into future sessions.

Install

Step 1 /plugin marketplace add nhevers/claude-recall Step 2 /plugin install claude-recall

How It Works

  1. Capture: Hooks into Claude Code lifecycle events (SessionStart, PostToolUse, Stop)

  2. Process: Worker service extracts observations from tool outputs

  3. Store: SQLite database with optional Chroma vector search

  4. Inject: Relevant context injected at session start via CLAUDE.md

Features

  • Automatic observation capture from tool outputs

  • Semantic search via MCP tools

  • Multi-provider support (Claude, Gemini, OpenRouter)

  • Web viewer UI for browsing history

  • Memory tagging and filtering

  • Export to markdown

  • Auto-pruning of old memories

Configuration

Settings stored in ~/.claude-recall/settings.json:

Setting

Default

Description

CLAUDE_RECALL_WORKER_PORT

37777

Worker service port

CLAUDE_RECALL_CONTEXT_OBSERVATIONS

50

Max observations to inject

CLAUDE_RECALL_PROVIDER

claude

AI provider for summaries

CLAUDE_RECALL_PRUNE_DAYS

0

Auto-prune after N days (0=disabled)

CLI Commands

claude-recall stats # Show memory statistics claude-recall export # Export to markdown claude-recall tag <id> <t> # Tag an observation claude-recall prune # Manual cleanup

Requirements

  • Node.js 18+ or Bun 1.0+

  • Claude Code

License

AGPL-3.0

-
security - not tested
F
license - not found
-
quality - not tested

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/nhevers/claude-recall'

If you have feedback or need assistance with the MCP directory API, please join our Discord server