Recall
Recall is a memory-as-a-service system for Claude Code and other Model Context Protocol clients. It is built to give Claude a permanent memory store that survives session restarts and context compaction, so earlier decisions, patterns, preferences, and tasks can be retrieved in later sessions.
The product’s core behavior is automatic memory capture and restoration through Claude Code hooks. According to the documentation, install the plugin once and four hooks handle context capture and restore on every session with no prompting required. Recall also supports cross-session workflows with named threads, automatic consolidation of similar memories while keeping originals as history, persistent to-do lists with priority levels, and semantic search based on vector embeddings and cosine similarity. It organizes memories by types such as decisions, patterns, preferences, and errors, and it includes real-time knowledge sharing so one Claude can learn something and other instances can know it immediately.
Recall is described for developers who want AI agents to remember conversations, learn patterns, and maintain context across sessions, projects, and teams. The page gives examples that include multi-session continuity, deployment monitoring, cross-session bug hunts, automatic decision records, multi-project switching, and sprint tracking. It also states that webhook ingestion can receive events from GitHub, Stripe, Sentry, Slack, Linear, PagerDuty, Discord, or any custom system and route them into Claude sessions, with HMAC signature verification, event history, replay, and audit trails.
Delivery is through a Claude Code plugin, MCP support for Claude Desktop and other MCP clients, and a REST API with CRUD operations, search, and batch processing. The MCP configuration shown uses the URL https://recallmcp.com/mcp. Pricing is shown as free on the site, and the documentation mentions a Get Started Free option. The page also states that it is SOC 2 ready, uses AES-256-GCM encryption at rest with per-tenant keys derived via HKDF, supports complete data isolation per API key, and offers seven embedding providers: Voyage AI, Cohere, OpenAI, Deepseek, Grok, Anthropic, and self-hosted Ollama.
FreemiumMITCLIAPICloud-managed
168stars
19forks
6features
2025since
Overview
6 featuresIn the RAG, search & retrieval space, Recall takes a focused approach. It focuses on allowing AI agents to retain knowledge and context across sessions for improved automation and collaboration. Recall is a B2B product aimed at ai developers. There is a free tier, and paid plans start at $10. It runs on the command line and API.
Recall first shipped in 2025. Development happens publicly on GitHub with 168 stars and 6 commits in the last 90 days. Among its 6 catalogued features are persistent memory, semantic search, and MCP protocol support. It exposes integrations via an MCP server and a public API.
Summary written by a language model from the project’s public pages.
- Persistent memory
- Semantic search
- MCP protocol support
- Multi-tenant isolation
- Webhooks
- Team collaboration
Tags
persistent-memorysemantic-searchmcp-protocolai-agentsmemory-api
AI capabilities
Structured
Inference: Cloud API
Built with & integrations
Framework
Next.js
Hosting
CloudflareRailway
Auth
firebase_authsupabase_auth
Database
Firebase
AI providers
anthropiclocal_oss
Connectors
MCPAPI
Runs on
CLIAPI-onlyCloud-managed
Detected from
- Next.js
- /_next/static/ in the HTML · __next_f in the HTML
- Railway
- x-railway-edge header · x-railway-request-id header
- local_oss
- bollama in the HTML
- Cloudflare
- cf-ray header · cf-cache-status header
Trust & compliance
License
MIT
Verified signals
Indexing history
1What PulseGate has recorded for this listing
Frequently asked questions about Recall
- What is Recall?
- Recall focuses on allowing AI agents to retain knowledge and context across sessions for improved automation and collaboration. It is catalogued under RAG, search & retrieval on PulseGate.
- Who is Recall for?
- Recall is a B2B product built for ai developers.
- How much does Recall cost?
- Paid plans start at $10, and a free tier is available.
- What platforms does Recall run on?
- Recall runs on the command line and API.
- Is Recall still active?
- The GitHub repository shows 6 commits in the last 90 days.
- What are alternatives to Recall?
- Similar projects tracked by PulseGate include Adaptive Recall, Recall™, and AgentRecall.Adaptive RecallRecall™AgentRecall
- When did Recall launch?
- Recall first shipped in 2025.
- Is Recall open source?
- Recall has a public GitHub repository.
Similar projects
Closest matches by what these projects do