orchestrator
diapason.agents.hybrid.skillorchestra.orchestrator
¶
The SkillOrchestra eval-orchestrator loop.
Faithful port of orchestration/eval_frames.py:run_single — the
multi-round search -> reasoning -> answer ReAct loop. Each round:
- Build a context string from accumulated docs / code results / attempts.
- Ask the orchestrator model (with the 3 tools) for the next stage. The
prompt is the verbatim
build_skill_orchestrator_promptwhen a handbook is loaded, else the baseline"Problem: ... Choose an appropriate tool."string. - Parse the tool call + any
<skill_analysis>block; route the worker model alias through the configuredRoutingStrategy. - Execute the tool.
answerends the loop; the last round force-callsanswer.
The orchestrator step does raw SDK calls (Anthropic / OpenAI) because it
needs the parsed tool_use blocks back — the same thing
extract_response_content_and_tool_calls does in the original.
run_orchestrator(agent, problem, *, cfg, handbook, strategy)
¶
Run the eval orchestrator on one problem. Returns (answer, metadata).