The agent that never forgets.
And never waits to be asked.

Build agents that understand member identity, interests, relationships, and intent across your community's real activity.

Try the Agent Community demo View source on GitHub

Company

Legal Engineering Support Finance Sales

Personal

Leadership Market

Public

Agent Community

Cron

Every 15 min — matches new launches and posts to members

Match community members to new launches and posts in their space, and draft intro messages for the best two matches.

Trace

watcher.poll(since: last_run)

→ 3 events collected

2:00Sarahasked for two strong intros in embeddings infra this week

2:06GitHubDevon shipped a new embeddings eval harness

2:12Miraposted a fresh MCP auth breakdown to the community feed

github.search.users(topic: embeddings)

→ 42 recent contributors

recall(member: Sarah, topic: needs)

→ needs infra feedback · embeddings, MCP

recall(members, overlap: embeddings)

→ Devon Lin · Mira Sato

link(Sarah ↔ Devon)

→ match · shared topic: embeddings

intros.draft(sarah-devon)

→ draft ready · references Devon's repo

Skills

Learn about Skills →

MCP

githubintros

Network

github.com.github.com.githubusercontent.comregistry.npmjs.org.npmjs.org

Memory

Learn about Memory →

Entities touched

MemberEntity: Sarah Chen

primarymember

CompanyCompany: Relay Labs

companycontext

ProjectProject: Relay Labs platform

projectactive

RepositoryRepository: eval-orchestrator

reposignal

PostPost: Why agent memory needs structure

postcontent

MatchMatch: Priya Natarajan

derivedintro

Relationship

member Sarah Chenworks_atcompany Relay Labs

Outcome →

#community-matches

Top matches for Sarah this week are Devon Lin (shipped a similar embeddings eval harness) and Mira Sato (deep MCP work). Intro drafts for both are queued in your outbox referencing Devon's repo and Mira's recent post.

Available in

Slack Telegram WhatsApp Discord Teams Google Chat REST API

Live workspace

Open workspace →

Read-only preview of the Agent Community workspace running on Lobu.

Architecture

Every agent runs in an isolated embedded worker with gateway-mediated network access — on your infrastructure, not ours.

Messaging platforms

Slack

Block Kit, interactive actions

Mini App, inline buttons

Reply buttons, list menus

Discord

Servers, DMs, markdown replies

Teams

Channels, bots, enterprise workflows

Google Chat

Cards v2, Workspace spaces

Link users across platforms with single sign-on
Approval flows, rich cards, buttons, and more

Bring your own agent

any MCP-capable agent

Lobu Memory

Lobu

Control Plane

Secrets

Single Sign-On (IdP)

Skill Registry Traces Sandboxing

Workers never see secrets
HTTP proxy with domain allowlist
MCP proxy with per-user OAuth
BYO provider keys (Anthropic etc.)

Equip your agent

Lobu Skills

OpenClaw Runtime

User A

isolated

OpenClaw Runtime

User B

isolated

OpenClaw Runtime

User C

isolated

One sandbox per user and channel
Subprocess isolation with just-bash virtual filesystems
systemd-run hardening on Linux production hosts
No direct internet access (gateway proxy only)
Nix reproducible environments
OpenTelemetry for observability

Lobu in memory benchmarks

Apples-to-apples comparison on public memory datasets. The model is GLM 5.1 via z.ai.

LongMemEval (oracle-50)

Single-session knowledge retention.

System	Overall	Answer	Retrieval	Latency
Lobu	87.1%	78.0%	100.0%	237ms
Supermemory	69.1%	56.0%	96.6%	702ms
Mem0	65.7%	54.0%	85.3%	753ms

LoCoMo-50

Multi-session conversational memory.

System	Overall	Answer	Retrieval	Latency
Lobu	57.8%	38.0%	79.5%	121ms
Mem0	41.5%	28.0%	66.9%	606ms
Supermemory	23.2%	14.0%	36.5%	532ms

Read the methodology

From the blog

Latest blog posts

memoryApr 13

Filesystem vs Database for Agent Memory

Agents need a workspace to think in and a warehouse to remember in. The filesystem is for ephemeral work. The memory layer is for durable organizational knowledge.

Read post →

mcpApr 2

MCP Is Overengineered, Skills Are Too Primitive

MCP HTTP is great for external services. MCP stdio is redundant. And most skill systems are just prompt text with no reproducibility. Here's what we built instead.

Read post →

announcementMar 2