# SuperOptiX

> Agent-to-agent communication, without framework lock-in. Make the agents you
> already run callable by any other agent, and get them discovered.
>
> SuperOptiX is the Agent-to-Agent (A2A) interoperability and optimization layer,
> built by Superagentic AI. It gives an agent you already run an A2A interface
> without a rewrite, then measures and improves whether other agents discover and
> call it. It records zero failures on the official A2A Technology Compatibility
> Kit: every requirement the TCK exercises passes, 73 of 73 at MUST.

SuperOptiX solves two problems in agent-to-agent communication. First,
interoperability: agents built on different frameworks cannot call each other,
and A2A (Agent2Agent) is the protocol that lets them. `super a2a adapt` reads an
agent you already wrote, derives the skills a calling agent would route on, and
writes an Agent Card plus a conformant A2A server. Your agent code is not
modified. Second, discoverability: being reachable is only half the problem,
because whether another agent chooses to call yours depends on how your Agent
Card describes it. SuperOptiX scores discovery rate, invocation rate and
confusion rate over a query set, then uses GEPA to rewrite the skill
descriptions against that score.

Eight agent runtimes are supported for A2A adaptation, agent registration and
agent orchestration: DSPy, CrewAI, OpenAI Agents SDK, Pydantic AI, Google ADK,
Claude Agent SDK, DeepAgents and Microsoft Agent Framework. Adaptation works on
agents written directly against those libraries, whether or not SuperOptiX built
them.

Install: `curl -fsSL https://superoptix.ai/install.sh | sh` (or `uv tool install superoptix`).
License: MIT. Vendor: Superagentic AI.

## Live A2A agent

- [A2A endpoint](https://a2a.superoptix.ai): a live SuperOptiX A2A agent on Google Cloud Run, serving A2A 1.0 and A2A 0.3 from one endpoint over JSON-RPC 2.0 and HTTP+JSON
- [Agent Card](https://superoptix.ai/.well-known/agent-card.json): the published A2A Agent Card describing the agent's skills, protocol bindings and supported interfaces

## Start here

- [SuperOptiX homepage](https://superoptix.ai/): what the A2A interoperability and optimization layer does
- [A2A support](https://superoptix.ai/a2a): agent-to-agent protocol support, conformance results and the live endpoint
- [Quick start](https://docs.superoptix.ai/quick-start/): adapt an agent to A2A in five minutes
- [Adapt an existing agent](https://docs.superoptix.ai/guides/a2a-adapt/): one command turns an existing agent into an A2A service
- [A2A conformance](https://docs.superoptix.ai/guides/a2a-conformance/): zero failures against the official A2A TCK, version negotiation, error binding, Agent Card caching
- [Routing quality](https://docs.superoptix.ai/guides/a2a-routing/): measuring and improving agent discoverability with GEPA
- [A2A introduction](https://docs.superoptix.ai/guides/a2a-introduction/): what the Agent2Agent protocol is and why an agent needs one

## Agent frameworks supported

- [All frameworks](https://superoptix.ai/frameworks): the eight supported agent runtimes
- [DSPy](https://superoptix.ai/frameworks/dspy)
- [OpenAI Agents SDK](https://superoptix.ai/frameworks/openai)
- [CrewAI](https://superoptix.ai/frameworks/crewai)
- [Google ADK](https://superoptix.ai/frameworks/google-adk)
- [Pydantic AI](https://superoptix.ai/frameworks/pydantic-ai)
- [Claude Agent SDK](https://superoptix.ai/frameworks/claude-sdk)
- [DeepAgents](https://superoptix.ai/frameworks/deepagents)
- [Microsoft Agent Framework](https://superoptix.ai/frameworks/microsoft)

## Agent optimization

- [Optimization](https://superoptix.ai/optimize): improving agent behaviour and agent routing
- [GEPA](https://superoptix.ai/gepa): the Genetic-Pareto optimizer used for Agent Card routing quality
- [GEPA guide](https://docs.superoptix.ai/guides/gepa-optimization/)

## Agent management and orchestration

- [Orchestration](https://superoptix.ai/orchestration): multi-agent orchestration and agent coordination
- [Marketplace](https://superoptix.ai/marketplace): 193 pre-built agents across 25 industries, agent discovery and installation
- [SuperSpec](https://superoptix.ai/superspec): the declarative YAML format agents compile from
- [CLI](https://superoptix.ai/super-cli): the `super` command line for agent management
- [Memory](https://superoptix.ai/memory) and [vector databases](https://superoptix.ai/vector-databases)
- [Observability](https://superoptix.ai/observability): agent tracing and monitoring

## Reference

- [Documentation](https://docs.superoptix.ai/): the full documentation site
- [All guides](https://docs.superoptix.ai/guides/): 67 guides
- [Source on GitHub](https://github.com/SuperagenticAI/superoptix)
- [PyPI package](https://pypi.org/project/superoptix/)
- [Superagentic AI](https://super-agentic.ai/): the company behind SuperOptiX

## Optional

- [Full corpus for agents](https://superoptix.ai/llms-full.txt): the same material with more detail
- [Tiers](https://superoptix.ai/tiers): engagement scope with Superagentic AI, separate from the MIT-licensed package
- [About](https://superoptix.ai/about) and [contact](https://superoptix.ai/contact)