orchestratelisted

# Orchestrate <p align="center"><img src="assets/architecture.svg" alt="orchestrate (Codex): an active GPT-6 Astra or GPT-5.6 Sol session leads; Astra keeps the hardest reasoning in the lead, Sol escalates unusually difficult units to Astra, and both route bounded work to lower tiers and can reach a Fable 5.1 cross-vendor peer" width="900"></p> ## Preflight the lead runtime first Before reading the task brief, inspecting the workspace, planning, or delegating, resolve `SKILL_DIR` as the directory containing this `SKILL.md` and run: ```bash "$SKILL_DIR/scripts/check-lead-runtime.sh" ``` This verifies the current thread through `CODEX_THREAD_ID`, its latest `turn_context`, and any recorded reroute. Proceed when it reports `gpt-6-astra` or `gpt-5.6-sol` and the actual effort. Preserve the user's selected effort; `xhigh` is not required. Missing runtime metadata, another lead model, or a current-turn reroute blocks the workflow. Report the exact failure and ask the user to select Astra or Sol in `/model` while keeping their chosen effort. Do not infer the current runtime from configuration defaults, the newest session file, or model self-identification. Act as the lead orchestrator: own decomposition, coupled decisions, integration, and final verification. Use the mode reported by the gate: - **Astra-lead mode:** keep compact hard reasoning and judgment in the lead. Delegate for width, isolation, execution, or an independent line. - **Sol-lead mode:** keep planning, coordi