Bridge Dots to a running Claude Code instance
The poster built a ChatGPT MCP so Dots could read and write to their running Claude Code instance in a multi-lane coding workflow, but found reliability issues. ·
Done withDotsChatGPTClaude Code
- What they did
- A commenter said they were already running many orchestrator and subagent coding lanes in Claude Code desktop, then built a quick ChatGPT MCP so Dots could read from and send messages into that live Claude Code setup. They also kept Codex open and used Dots as the main long-running session that collected updates and let them talk across those lanes with less back-and-forth.
- How it went
- They said it took a few hours before Dots started giving good results, and that with setup work it could become impressive. The main limitation they called out was reliability issues, plus confusion over whether Dots was actually writing versus just creating or messaging Claude Code sessions.
- Worth knowing
- Their main practical warning was that Dots was bad out of the box and only became useful after a few hours of tuning.
Try it yourself with Dots
I want to run several coding lanes, with [another agent] handing work to this Claude Code session and reading back results. Set up a simple, reliable bridge, such as an MCP server or a shared task file, for [project folder]. Test that messages go both ways, log any failures, and show me the setup before connecting anything else.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: Reddit · Oct 1, 2026
Related use cases
Turn tickets into merged pull requests
Pulls tasks from GitHub Issues or Linear, then runs Claude Code or Codex in isolated Kubernetes pods to carry each one through to a PR.
Run ten business tasks through an always-on agent
A reviewer tested Dots on product research, publishing a webpage, content planning and scheduled jobs; inbox triage and follow-ups needed supervision and a spreadsheet had broken formulas.
Track incoming emails and notify about key ones
The user asked Dots to watch for certain incoming emails and notify them, but it did not do so and was very slow.
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 Terraform plan and apply through an MCP agent
The author had Cursor run Terraform against Postgres via MCP; the apply hit a permissions error and he finished from the terminal.
Scaffold a second module from an existing one's structure
A commenter asked Cursor to create a second module following another module's structure, but it failed because it did not read the relevant files.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.