AgentFlow Docs

Build multi-agent AI systems that actually ship. Agentflow is the production-grade framework that takes you from prototype to production without rewriting your stack.

Most agent frameworks stop at a demo. Agentflow ships the whole thing — orchestration, memory, API, auth, SDK, and UI — so you spend your time on agents, not plumbing.

• 🧠 Memory that remembers — a 3-layer system: working state → Redis hot cache → Postgres durable → vector recall (Qdrant/Mem0). Your agents have short-term speed and long-term knowledge.
• 🔌 Any model, no lock-in — OpenAI, Google GenAI, Anthropic, or your own. Swap providers with a config change, not a rewrite.
• ⚡ Live agents, real-time — token-by-token streaming, live state inspection, and parallel tool execution by default.
• 🕸️ Graph-based orchestration — model complex, cyclic agent workflows with StateGraph. LangGraph power, far less ceremony.
• 🛠️ Production from day one — auto-generated FastAPI backend, JWT/RBAC auth, rate limiting, dual-layer checkpointing, and Docker/Kubernetes builds.
• 🤝 MCP native — plug into the Model Context Protocol ecosystem out of the box.
• 📦 Batteries included — backend, REST API + CLI, typed TypeScript SDK, and a visual React playground. One framework, end to end.

From pip install to a streaming multi-agent API in minutes.

This repo holds the documentation for that ecosystem.

📖 Read the docs: https://agentflow.10xscale.ai

Built with Docusaurus 3. Deployed to agentflow.10xscale.ai via GitHub Pages.

🧩 The Agentflow Ecosystem

Package	What it does	Install
Core framework (10xscale-agentflow)	Graph-based agent orchestration, 3-layer memory, parallel tools, MCP	pip install 10xscale-agentflow
API + CLI (10xscale-agentflow-cli)	FastAPI server auto-generated from your graph, auth, RBAC, rate limiting	pip install 10xscale-agentflow-cli
Client SDK (@10xscale/agentflow-client)	Typed TypeScript/React client with streaming hooks	npm install @10xscale/agentflow-client
Playground	Visual React UI to test agents against a local server	agentflow play
Docs (this repo)	Tutorials, how-to guides, reference, and concepts	https://agentflow.10xscale.ai

Local development

npm install
npm run start

Dev server: http://localhost:3000.

Build

npm run build
npm run serve

Static output in build/.

Windows + Git Bash note: if npm run build mangles BASE_URL=/ (you'll see broken links resolving to C:/Program Files/Git/...), run from PowerShell with $env:MSYS_NO_PATHCONV='1'.

Checks

npm run build              # fails on any broken internal link
npm run typecheck
npm run lint:frontmatter   # every page needs title, description, keywords
npm run lint:links         # every external URL must resolve
npm run verify:api         # documented symbols, routes, and commands must exist
npm run og-image           # regenerate the PNG social card from the SVG

verify:api checks the docs against the packages published to PyPI, which is what readers actually install:

pip install 10xscale-agentflow 10xscale-agentflow-cli
npm run verify:api

All of these run in CI (.github/workflows/ci.yml).

Deploy

.github/workflows/deploy.yml builds and deploys to GitHub Pages on push to main.

Required GitHub repo secrets (optional but recommended):

• GOOGLE_ANALYTICS_ID — e.g. G-XXXXXXXXXX
• MICROSOFT_CLARITY_ID — for heatmaps / session replay

The site is configured for the custom domain agentflow.10xscale.ai (CNAME in static/).

Repo layout

docs/
  get-started/         # golden path, beginner-friendly
  beginner/            # guided tutorial path
  concepts/            # mental models, with an "In depth" tier beneath
  prebuild/            # prebuilt agents and tools
  how-to/              # task-oriented guides (python, production, cli, client)
  qa/                  # unit testing and evaluation
  tutorials/           # from-examples deep dives
  reference/           # API reference (Python, REST, CLI, TS client)
  troubleshooting/
  use-cases/           # production reference architectures
  integrations/        # FastAPI / Next.js / Postgres
  providers/           # LLM provider configuration
  glossary/            # definition pages
  compare/             # framework comparisons (LangGraph, CrewAI, AutoGen, etc.)
  courses/             # GenAI beginner + advanced curriculum
  project/             # changelog, upgrade guide, roadmap, security, support

blog/                  # cornerstone posts, RSS at /blog/rss.xml
src/
  components/          # CompareTable, FAQ, RelatedDocs, BlogStructuredData
  pages/               # Homepage
  theme/               # MDXComponents, Root swizzles
static/                # CNAME, robots.txt, llms.txt, social card, favicon
scripts/               # front-matter, link, and API-drift checks
COURSE_STYLE_GUIDE.md  # authoring rules for docs/courses (not published)

Docs versions are cut only when a release line needs to stay available:

npm run docs:cut-version -- 1.0

That snapshots docs/ into versioned_docs/version-1.0; the navbar version dropdown then appears automatically.

Contributing

PRs welcome. Before opening one, run the checks above and read CONTRIBUTING.md.

Writing conventions, where a page belongs, and the release process are documented on the site: Contributing.

Two rules worth repeating here:

• Verify before asserting. Read the source for the signature, the default, and the error message. Documented APIs that never existed have shipped before; npm run verify:api exists to stop that.
• Moving or renaming a page requires a redirect in docusaurus.config.ts.

Related repos

• 10xHub/Agentflow — the core Python framework (10xscale-agentflow)
• 10xscale-agentflow-cli — FastAPI server + CLI
• @10xscale/agentflow-client — TypeScript/React client SDK
• Playground — visual React UI for testing agents (agentflow play)
• Docs: https://agentflow.10xscale.ai

License

Agentflow is MIT licensed and made by 10xScale. Contributions are accepted under the same license.