Skip to main content

Open-source agent-memory MCP server. recall.works

Project description

Recall™

Open-source memory for AI agents. MCP-native. Self-hosted. One Docker image.

Tests Docker PyPI npm License: MIT Python 3.11+ MCP Container

OSS quickstart · Recall Pro → · Book a demo · IceWhisperer for Encompass

Your agent forgets every session. Recall fixes that — with a small, opinionated memory surface that any MCP-speaking agent (or any HTTP client) can drive. Append-only, rebuildable, soft-delete by design.

   ┌─────────────┐    MCP / HTTP    ┌──────────────────────────┐
   │  AI agent   │ ───────────────► │  Recall (one container)  │
   │  (Copilot,  │                  │   • 13 memory tools      │
   │   Claude,   │  remember/recall │   • BYO embedder + LLM   │
   │   Cursor,   │ ◄─────────────── │   • Append-only artifacts│
   │   custom)   │                  │   • Auto-snapshot to disk│
   └─────────────┘                  └──────────────────────────┘

Five-minute install

1. Run the server:

docker run -d --name recall \
  -p 8787:8787 \
  -e API_KEY=changeme \
  -v recall-data:/data \
  ghcr.io/recallworks/recall:latest

2. Talk to it — pick your stack:

# Raw HTTP (any language)
curl -H "X-API-Key: changeme" \
     -H "Content-Type: application/json" \
     -d '{"content":"first memory","tags":"hello"}' \
     http://localhost:8787/tool/remember
# Python (use requests/httpx — no SDK pkg needed)
import requests
h = {"X-API-Key": "changeme", "Content-Type": "application/json"}
requests.post("http://localhost:8787/tool/remember", headers=h,
              json={"content": "first memory", "tags": "hello"})
print(requests.post("http://localhost:8787/tool/recall", headers=h,
                    json={"query": "memory"}).json()["result"])
// TypeScript / JavaScript (Node 18+, Bun, Deno, browser)
npm install @recallworks/recall-client

import { RecallClient } from "@recallworks/recall-client";
const c = new RecallClient({ baseUrl: "http://localhost:8787", apiKey: "changeme" });
await c.remember("first memory", { tags: "hello" });
console.log((await c.recall("memory")).result);

Full walkthrough: docs/quickstart.md.


What you get

  • 13 toolsremember, recall, reflect, anti_pattern, checkpoint, pulse, session_close, index_file, reindex, snapshot_index, memory_stats, forget, maintenance.
  • Two transports — plain HTTP (POST /tool/{name}) and MCP over SSE. Drop into Copilot, Claude Code, Cursor, or any MCP client.
  • Bring your own models — pluggable embedder (default / OpenAI / Ollama) and summarizer (noop / OpenAI / Ollama). Run fully offline, fully on-prem, or against your own Azure-OpenAI tenant. See docs/byo-models.md.
  • Durable by default — ephemeral live store with auto-snapshot to disk; container restarts come up whole.
  • Append-only artifacts — every write also lands as a .md file. If the vector store ever burns down, reindex rebuilds it from the artifacts.
  • forget is soft-archive — guardrail wired into the OSS code itself, not bolted on as policy. Memory you delete can be recovered.

How it's different

Recall Mem0 / Letta / Zep
License (core) MIT mixed; SaaS-first
Self-host one docker run varies, often non-trivial
BYO embedder default / OpenAI / Ollama (env var) usually fixed
BYO LLM noop / OpenAI / Ollama (env var) usually fixed
Storage model append-only artifacts + vector index, rebuildable live DB only
delete soft-archive by design hard delete
Tool surface 13 opinionated tools (memory + workflow) embedding + retrieval primitives
MCP-native yes, plus plain HTTP partial / via wrapper
Ops model single binary, single container multi-service stack

If you want a managed service, see Recall Cloud below. If you want a brain you fully own, this OSS core is enough.


Repo layout

Path What
src/recall/ OSS server (MIT)
src/recall/tools/ One module per tool
src/recall/transport/ HTTP + MCP/SSE adapters
docker/single-tenant/ Reference Dockerfile + compose
tests/ pytest suite (no Docker required)
docs/ Quickstart, conventions, architecture
enterprise/ Multi-tenant, SSO, control plane (BSL)

Conventions

These are the practices that make the tools pay off. Pick what fits.

  • Cold-start ritual — opening protocol every session should run.
  • Branding — signed-edit headers so you can trace which agent touched which file when.

Status

Alpha. The code in src/recall/ is extracted from a hosted production brain that has served thousands of sessions, then sanitized of org-specific paths, extensions, and tenant data. Expect breaking changes before 1.0; pin the image tag.


Contributing

Yes — please read CONTRIBUTING.md first. We accept bug fixes, new Store backends, doc improvements, and anti-pattern entries. We don't accept architectural rewrites without prior discussion.

Security issues: see SECURITY.md.


License

  • src/recall/, clients/, docker/single-tenant/, docs/, examples/MIT (LICENSE)
  • enterprise/BSL 1.1, 5-seat additional-use grant, converts to MIT after 3 years (LICENSE-COMMERCIAL.md)

Recall Open Source vs. Recall Pro vs. Hosted

Capability OSS (this repo) Recall Pro Recall Cloud
Single-tenant Docker image n/a (hosted)
13 memory tools, MCP + HTTP
BYO embedder + LLM
Append-only artifacts + auto-snapshot
Multi-tenant, SSO, RBAC
Audit log + retention policy
Cross-session entity graph
PII sanitization pipeline
Snapshot replication / DR
Vendor support + SLA community business hours 24×7
Hosted on our infra
Pricing free from $99/mo per node from $0.10 per 1k tools

Recall Pro ships from the enterprise/ tree under a Business Source License — source-available, 5-seat free Additional Use Grant, converts to MIT after 3 years. Buy a license and the enterprise/ modules light up alongside your OSS install.

Recall Cloud is the hosted multi-tenant version. Same tools, no infra. Reach out for early-access pricing.

➡️ Talk to sales: sales@recall.works · Book a 20-min walkthrough: https://recall.works/demo


Vertical builds powered by Recall

Recall is the engine. We ship turn-key vertical brains on top of it:

  • IceWhisperer — the memory + workflow brain for ICE Mortgage Technology / Encompass shops. Pre-loaded SDK index, settings recipes, plugin audits, drift detection. Pilots from $250/mo.

If you want a vertical brain for your industry, we'll build it. Email partners@recall.works.


Maintainers

Reach the maintainers at maintainers@recall.works. Issues and PRs welcome on GitHub.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ai_recallworks-0.3.4.tar.gz (19.7 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

ai_recallworks-0.3.4-py3-none-any.whl (51.1 kB view details)

Uploaded Python 3

File details

Details for the file ai_recallworks-0.3.4.tar.gz.

File metadata

  • Download URL: ai_recallworks-0.3.4.tar.gz
  • Upload date:
  • Size: 19.7 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.10

File hashes

Hashes for ai_recallworks-0.3.4.tar.gz
Algorithm Hash digest
SHA256 00bbadeabac9064d2c1233bbf2d04615ec371bf94f78c19cd282277f71a1f592
MD5 7b357eb6c0322631cfb557f3956e5081
BLAKE2b-256 92e86c641c58fe2570210dfd43ccfbdef6046712402b6ce08e3a4cfb474cb9dd

See more details on using hashes here.

File details

Details for the file ai_recallworks-0.3.4-py3-none-any.whl.

File metadata

  • Download URL: ai_recallworks-0.3.4-py3-none-any.whl
  • Upload date:
  • Size: 51.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.10

File hashes

Hashes for ai_recallworks-0.3.4-py3-none-any.whl
Algorithm Hash digest
SHA256 61a7ae95dffd8a724bf2883d152b660a5005aab05162533519f84f462fd0e146
MD5 1fa40a0cac86693ad4eefcd4280b22c2
BLAKE2b-256 11a79a8051345187d3c8550b1f83f3d81241eda029bb384bf0d276b158b8b80c

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page