Claude Code session orchestration with Maxxwell
Maxxwell coordinates parallel Claude and Codex sessions through an orchestrator that monitors goal checklists and brings important decisions back to the user.
news.ycombinator.com
Prompt to try
Editorial adaptation, not the source author’s exact prompt. Replace bracketed inputs and review tool permissions first.
Coordinate [coding sessions] toward [goal]. Maintain a checklist for each goal, track progress and blockers, and report completed work, decisions made and questions that need my input. Keep worker sessions accessible for direct review.
Reach people exploring AI agents
See placements →What happened
The founder reported merged PRs per prompt rising from 0.9 to 5 and revert rate falling from 0.56% to 0.20% over a month of use.
Reported by mys1 on Hacker NewsHow the agent did it
- Run parallel Claude Code sessions in terminal panes.
- Use an orchestrator to align workers and filter questions.
- Monitor per-goal checklists for adherence.
Where a person stepped in
The founder communicates with the orchestrator about decisions requiring input and can attach directly to worker sessions.
A person set the goal and reviewed the result.
Michael, a founder of Rindler, describes building Maxxwell after managing twelve Claude Code sessions in terminal panes became exhausting. His team had been developing browser agents and tried several coding tools before settling on parallel Claude Code sessions. That arrangement offered control, but answering questions across terminals made progress and blockers difficult to track.
The team created an orchestrator focused on keeping worker sessions aligned with their goals and filtering their questions. Michael says he now primarily messages that single session. It reports what has landed, decisions made on his behalf and matters requiring his input. Workers remain unmodified Claude or Codex processes running in real PTYs, and a person can attach directly to any session.
Per-goal checklists give the orchestrator something to monitor for adherence. The founder reports using the system for hours at a time and for overnight runs. Over the month described, merged PRs per prompt at similar semantic weight rose from 0.9 to 5, while the reported revert rate fell from 0.56% to 0.20%. These are the founder’s measurements of their own usage.
Who it's for: Developers already juggling several coding sessions may benefit from a central place to review progress and intervene.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Hacker News · mys1
Frequently asked questions
How does the user supervise the workers?
The founder primarily messages one orchestrator, which reports completed work, decisions and questions. Users can also attach directly to any worker session.
What runs inside each worker session?
The source says each worker is an unmodified Claude or Codex process running in a real PTY.
More details
Remaining steps
- Review orchestrator reports and attach to workers when needed.
About this account
An orchestrator manages worker alignment and filters questions, with the founder reporting hours-long and overnight runs.
Platform: Hacker News. Attribution: https://news.ycombinator.com/item?id=49625975. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: mys1 on Hacker News
Source published 9 Sept 2026
Source checked 14 Oct 2026
Tools and topics
Claude Code · Codex · Maxxwell
Orchestration · Parallel agents · Code review
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.