Skip to content

Runtimes and harnesses

Runtime adapters execute a HarnessSpec through another agent framework while preserving the SuperQode contract and normalized evidence where the adapter supports it.

Runtime Optional dependency Select or run Verify Detailed guide
Google Agent Development Kit uv tool install "superqode[adk]" runtime.backend: adk superqode runtime doctor adk Python Runtimes
OpenAI Agents SDK uv tool install "superqode[openai-agents]" runtime.backend: openai-agents superqode runtime doctor openai-agents Python Runtimes
PydanticAI uv tool install "superqode[pydanticai]" runtime.backend: pydanticai superqode runtime doctor pydanticai Python Runtimes
PydanticAI with Logfire uv tool install "superqode[pydanticai-logfire]" Enable PydanticAI tracing in the HarnessSpec superqode runtime doctor pydanticai Agent Runtimes
DeepAgents uv tool install "superqode[deepagents]" runtime.backend: deepagents superqode runtime doctor deepagents Agent Runtimes
RLM Code uv tool install "superqode[rlm-code]" runtime.backend: rlm-code superqode harness protocol describe rlm-code RLM Code
Hugging Face Tau uv tool install "superqode[tau]" :tau use or :harness switch tau :tau status Hugging Face Tau

Omnigent compatibility

Omnigent support is an included format and execution bridge. It imports a compatible agent.yaml into a repository-owned HarnessSpec; it does not run the Omnigent server or managed host workflow.

superqode harness import-omnigent path/to/agent.yaml --output harness.yaml
superqode harness validate --spec harness.yaml
superqode harness doctor --spec harness.yaml

See Omnigent Compatibility for field mapping and the interoperability boundary.