Orchestrate multiple coding agent sessions from one Claude session
A developer talks to one Claude session that messages other agent sessions (including Codex) as orchestrator, reachable from CLI, mobile app and IRC, to load-balance token usage. ·
Done withClaudeClaude CodeCodex
- What they did
- The developer talks mostly to a single Opus 5.5 session, chosen for its 1M context window, and lets it message other agent sessions as an orchestrator. They use it from the CLI and the mobile app. Claude often calls Codex on their behalf. Some agents are also reachable over IRC through a small CLI they're building.
- How it went
- The workflow replaced the need for complex orchestration tools or agent platforms. Delegation was once for quality problems; now it mainly load-balances token usage across Claude and Codex subscriptions. The IRC CLI is still in progress.
- Worth knowing
- This assumes subscriptions to both Claude and Codex, mostly at top tiers, which the author adjusts up or down depending on usage.
Try it yourself with Claude
Help me set up an orchestrator workflow where this Claude Code session can send tasks to other agent sessions, including [Codex or another coding agent], and collect their replies. Start by proposing a simple setup for [my project folder] where I give you one goal and you split it into [number] parallel tasks. Show me the plan and wait for my approval before launching any sessions, and finish by summarizing what each session did.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: HN · Nov 4, 2026
Related use cases
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Run side hustle invoicing, payments and bookkeeping
A side hustler connected their bank to Claude via MCP and QuickBooks so an agent handles invoicing, contractor payments, expense tracking and bookkeeping.
Split frontend and backend work across Claude Code and Codex
A poster ran Claude Code and Codex in parallel for 15 days, with Claude Code on the frontend and another agent on the backend.
Review Linux kernel patches
Google engineers launched Sashiko, an agentic AI code reviewer for the Linux kernel.
Custom agent
Build a budgeting app from a single prompt
A low-coder used Replit Agent to build a 'Budget Buddy' with category budgets, a spending dashboard and expense log in about 20 minutes, using half their monthly credits.
Build a project management tool with a 24-hour agent run
Ran a Cursor agent continuously for 24 hours to build a project management tool; the app fully works but is not production ready.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.