{"slug": "practical-multi-agent-orchestration-in-codex", "title": "Practical multi-agent orchestration in Codex", "summary": "OpenAI's Codex introduces Multi-Agent V2 tools, enabling GPT-5.6 Sol and Terra to delegate tasks, share updates, and coordinate complex work, with concurrency defaulting to four agents. The system supports role-based agents (Scout, Worker, Smart Worker) with adjustable reasoning effort and context inheritance, and recommends capturing orchestration patterns in skills for efficiency.", "body_md": "GPT-5.6 Sol gets especially interesting when it has a team to work with. Codex's new Multi-Agent V2 tools give Sol and Terra a natural way to delegate tasks, share updates, and coordinate through complex tasks.\n\nUltra makes agent coordination the default and is best reserved for high-stakes work, where ambiguity or scattered context justify the added depth.\n\nFor other tasks, a short prompt or skill can encourage the same collaborative behavior from Sol Medium, as it stays in conversation with you while organizing the work behind the scenes. With the right nudge, Sol can turn broad requests into focused assignments, bringing in other agents, and deciding when a problem calls for deeper reasoning.\n\nMatch reasoning to the work\n\nWhile it's possible to have Sol delegate to another model like Terra, the simplest setup keeps one model family and adjusts only the reasoning effort, with dedicated roles like so:\n\nScout — GPT-5.6 Sol Light. Answer narrow, read-only questions: locate files, trace a code path, or find relevant tests.\n\nWorker — GPT-5.6 Sol Medium. Implement scoped changes, run checks, or handle supporting work.\n\nSmart worker — GPT-5.6 Sol High. Take on difficult implementation, resolve ambiguity, or coordinate help when useful.\n\nTreat these roles as useful defaults. Sol Light retains the judgment to find useful context without spending as much reasoning on discovery.\n\nLet the team coordinate\n\nThe coordinator acts as the primary delegator: it assigns substantive work, avoids duplicate investigations, and tracks what each agent is doing. Scouts can investigate in parallel, while workers can share implementation when responsibilities are clear.\n\nAgents can also message one another directly through a common messaging system with separate inboxes. When a scout discovers something a worker needs, it can recognize the dependency and pass along its findings without waiting for the coordinator to relay them.\n\nConcurrency is configurable per thread and defaults to four agents, including the coordinator. Within that budget, a smart worker might coordinate a scout and another worker, or the coordinator could send three scouts to investigate separate questions.\n\nChoose what context agents inherit\n\nForking conversation history helps agents understand the broader goal and earlier decisions. Starting with fork_turns: \"none\" gives an agent a fresh, focused assignment instead. Fresh-context agents can still recognize when a teammate needs information and contact them independently.\n\nAgents that inherit their parent's context may also see its orchestration instructions. When an agent should remain a leaf, give it a short boundary:\n\nComplete this assignment directly. Do not spawn other agents; your parent's delegation instructions apply only to your parent.\n\nFresh-context agents won't inherit task-specific tool or safety boundaries, so include any essential restrictions directly in their assignments.\n\nCapture the pattern in a skill\n\nA practical skill can give the coordinator a few standing instructions:\n\nStay available to the user while delegating substantive work. Send focused, read-only scouts out in parallel with reasoning_effort: \"low\" and fork_turns: \"none\". Use reasoning_effort: \"medium\" for routine implementation and reasoning_effort: \"high\" for harder problems. Give each agent clear ownership, avoid overlapping assignments, and tell leaf workers not to delegate. Bring the results together and keep approvals with the user.\n\nExperiment with the knobs\n\nStart with these defaults, then experiment with reasoning effort, context inheritance, delegation authority, and how agents collaborate. The goal is to understand which settings help a team move the work forward without spending more reasoning than the task requires.", "url": "https://wpnews.pro/news/practical-multi-agent-orchestration-in-codex", "canonical_source": "https://twitter.com/pvncher/status/2080707291603407077", "published_at": "2026-08-17 15:54:17+00:00", "updated_at": "2026-08-17 16:11:40.055697+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "developer-tools"], "entities": ["OpenAI", "Codex", "GPT-5.6 Sol", "Terra"], "alternates": {"html": "https://wpnews.pro/news/practical-multi-agent-orchestration-in-codex", "markdown": "https://wpnews.pro/news/practical-multi-agent-orchestration-in-codex.md", "text": "https://wpnews.pro/news/practical-multi-agent-orchestration-in-codex.txt", "jsonld": "https://wpnews.pro/news/practical-multi-agent-orchestration-in-codex.jsonld"}}