Documentation

Everything you need to build, deploy, and scale AI agent workflows with SwarmAI.

Module Reference

SwarmAI is organized into 8 focused Maven modules

swarmai-core

Agents, tasks, processes, state, skills, memory, knowledge, budget, governance, observability. Includes the provider-neutral LlmClient SPI, lifecycle hooks, compliance gates, and the built-in RAG pipeline (hybrid retrieval + citations)

swarmai-tools

74 built-in tools: coding (build/test/lint/format/edit/diff/git/grep), research (Arxiv, Wikipedia, Wolfram), finance (DCF, LBO, comps, tear sheet), office (Xlsx/Pptx authoring + audit), web, S3, Kafka, OpenAPI

swarmai-dsl

YAML DSL parser & compiler for declarative workflow definitions

swarmai-native-anthropic

Native Anthropic adapter — prompt caching, extended/adaptive thinking, inline citations, server-side tools, Message Batches, Files API

swarmai-native-ollama

Native Ollama adapter for local-first inference with adaptive routing

swarmai-self-improving

10% community-investment loop: improvement collector, pattern extractor, GitHub PR reporting, anonymized telemetry export

swarmai-distributed

Cross-process swarms, shared skill registry, distributed coordination

swarmai-enterprise

Multi-tenancy, deep RL (DQN), advanced governance, RBAC, audit, billing (BSL 1.1)

swarmai-eval

Self-evaluation swarm, competitive benchmarks, value scoring, regression detection

swarmai-studio

Web dashboard for workflow monitoring, event replay, and debugging

swarmai-bom

Bill of Materials for dependency version alignment

Quick Start

<!-- Add to your pom.xml -->
<dependency>
    <groupId>ai.intelliswarm</groupId>
    <artifactId>swarmai-core</artifactId>
    <version>1.0.28</version>
</dependency>