In the example CXU, the PyranaHarness implements native LLM tool calling via LiteLLM by executing a Turn 0 plan followed by iterative tool-calling loops until the agent finalizes an AgentOutputEnvelope.

claim: "The PyranaHarness implements native LLM tool calling via LiteLLM,

— PYRANA_Internal_Knowledge_Graph_Specification.md

executing a Turn 0 plan followed by iterative tool calling loops

— PYRANA_Internal_Knowledge_Graph_Specification.md

until the agent finalizes an AgentOutputEnvelope"

— PYRANA_Internal_Knowledge_Graph_Specification.md

"PyranaHarness (pyrana_services/orchestration/harness/pyrana_harness.py) which implements native tool calling via LiteLLM"

— PYRANA_Internal_Knowledge_Graph_Specification.md

"Turn 0: create initial plan via LLM"

— PYRANA_Internal_Knowledge_Graph_Specification.md

"Finalize: build and persist AgentOutputEnvelope"

— PYRANA_Internal_Knowledge_Graph_Specification.md