Loma documentation

Loma is an AI Agents Factory for your whole team — one agent, in Slack and a dashboard, that knows your tools, runs on open models or your pooled Claude subscriptions, automates work on schedules and webhooks, and shares a team-wide skill library.

These docs cover installing, configuring, and operating your own Loma deployment. If you’re new, start with Getting started.

How Loma fits together

  • Backend — a Python service (aiohttp + Slack Bolt) that runs the agent, serves the API, and receives webhooks. Listens on port 3000.
  • Dashboard — a Next.js app for chat, conversation history, skills, flows, integrations, configuration, users, and usage. Listens on port 3001.
  • MongoDB — stores conversations, skills, flows, users, integrations, and usage.
  • Agent runtimeOpenCode by default (open-source models); optionally a pool of Claude Code accounts.
  • Skill assets — non-text skill files live on local disk at LOMA_SKILL_ASSET_DIR.

Where to go next

SectionWhat’s inside
Getting startedQuickstart, a fresh EC2/GCP install, and local development
ConfigurationEnvironment reference, feature flags, editing config from the dashboard
Agent runtimeOpenCode, Claude account pooling, choosing a model
SkillsAuthoring, importing, assets, and versioning team knowledge
Flows & automationsScheduled routines and webhook-triggered agent tasks
IntegrationsConnecting databases, issue trackers, CRM, docs, and observability
Slack appCreating and installing the Slack bot
AuthenticationLocal credentials and Google OAuth
Deployment & networkingReverse proxy, single-port exposure, HTTPS, backups
Security & rolesPosture and role-based access

Source code

Loma is open source at github.com/plotlinelabs/loma. This repository is the clean OSS codebase — company-specific knowledge, prompts, playbooks, and credentials belong in your database and environment, never in source.