cortex-agent-framework 1.0.0

Fan-Out/Fan-In agentic orchestration framework with multi-provider LLM support and MCP tool integration.

Cortex Agent Framework Build production-grade AI agents in a single YAML file. Fan-out / fan-in orchestration, multi-LLM routing, MCP tool servers, quality validation, and delta learning — all driven by cortex.yaml.

Overview • Architecture • Features • Getting Started • Use Cases • Config • CLI • Deployment • FAQ

---

Why Cortex?

Every production AI agent eventually needs the same plumbing: task decomposition, parallel tool execution, streaming, validation, retries, session management, multi-provider routing, MCP integration, deployment. Most teams rebuild this stack three times before they ship.

Cortex is that stack, pre-built and configuration-driven. Define your agent once in cortex.yaml, call framework.run_session(), and get a production-ready agent with parallel tool execution, quality scoring, session persistence, hot-reload, and deployability as a Docker service, Python package, or MCP server.

agent:
  name: ResearchAgent
  description: Searches the web and writes reports

llm_access:
  default:
    provider: anthropic
    model: claude-sonnet-4-5
    api_key_env_var: ANTHROPIC_API_KEY

task_types:
  - name: web_research
    capability_hint: web_search
    output_format: md
  - name: write_report
    capability_hint: document_generation
    depends_on: [web_research]

from cortex.framework import CortexFramework

framework = CortexFramework("cortex.yaml")
await framework.initialize()

result = await framework.run_session(
    user_id="user_1",
    request="Research the latest vector DB benchmarks and write a report",
)
print(result.response)

That's the whole agent. Fan-out, tool calls, dependency resolution, synthesis, validation — all handled.

---

What you get out of the box

8 LLM providers	Anthropic, OpenAI, Gemini, Grok, Mistral, DeepSeek, AWS Bedrock, Azure AI — swap with one YAML line
Fan-out / fan-in	Primary agent decomposes into a dependency graph, tasks run in parallel, results synthesised automatically
MCP-native	First-class support for SSE, stdio, and streamable-HTTP MCP tool servers
Multi-agent mesh	Publish any agent as an MCP server — other agents consume it as a tool. Compose specialist agents into an orchestrator
Quality validation	Every response scored on intent match, completeness, coherence. Enforced threshold floor
Delta learning	Agent proposes new task types from usage patterns; human-in-the-loop review before apply
Streaming	Typed event stream (StatusEvent, ResultEvent, ClarificationEvent) wires cleanly into FastAPI SSE
Session persistence	Memory / SQLite / Redis backends with WAL replay and resumable sessions
Dev ergonomics	Browser setup wizard, hot-reload, dry-run validation, session replay, config migration
Deployability	One command publishes as Docker image, Python wheel, or MCP server

---

Get started in 5 minutes

git clone <repo-url>
cd cortex-agent-framework
python -m venv .venv && source .venv/bin/activate
pip install -e .

export ANTHROPIC_API_KEY=sk-ant-...
cortex setup             # browser wizard at localhost:7799
cortex dry-run "Hello"   # validates config, no LLM calls
cortex dev               # run in hot-reload mode

Full walkthrough → docs/GETTING_STARTED.md

---

Documentation

Document	Read this if you want to…
Overview	…understand what Cortex is, who it's for, and why it exists
Architecture	…see the internals: primary agent, task graph compiler, MCP agents, validation, learning
Features	…scan the full feature matrix before committing
Getting Started	…build your first agent with working code examples
Use Cases	…see real-world scenarios and reference architectures
Configuration	…look up every cortex.yaml field
CLI Reference	…look up every cortex subcommand
Deployment	…ship to production (Docker, wheel, MCP, multi-agent mesh)
FAQ	…find answers to common questions and gotchas
Contributing	…report bugs or submit pull requests

---

Community & support

• Issues: file bugs and feature requests on GitHub Issues
• Discussions: ask questions and share configs on GitHub Discussions
• Security: please report vulnerabilities privately, not in public issues

---

License

MIT — see LICENSE. Use it in commercial products, fork it, modify it, ship it.