Image: GitHub
Why it mattersA team pays per token for a coding agent on each issue. Grouping that work behind one review pipeline is how the cost stops growing per seat.
Superplane released version 1.0 of its open-source tool for running coding agents as a team pipeline. The company blog post of 27 October announces the release. The GitHub repository was at 9,016 stars this morning. It ranked 16th on Trendshift's daily list of trending repositories.
The tool coordinates one or more coding agents. The work path goes from a backlog ticket to a merged pull request. A reader can self-host the tool. One option is a single Linux virtual machine with Docker Compose. Another is Kubernetes on AWS, Azure or Google Cloud. The source is on GitHub under the Apache-2.0 licence.
What the pipeline does
Superplane pulls work from GitHub, Jira, Linear or Sentry, the company says on its site. Each task starts a coding agent in its own cloud development environment. Many agents can run in parallel. They do not touch each other's files. The agent writes the code and opens a pull request. The pipeline runs CI. A reviewer may leave a comment. The agent then reads the feedback and tries again, the blog post says. A person decides what gets merged.
The company says the tool works with "OpenAI, Anthropic, Google, xAI, DeepSeek, Qwen and Kimi". It does not name the agent frameworks it can drive. The README lists GitHub as its one explicit source-control integration.
What the numbers come from
The 9,016-star figure is the live GitHub count. The repository was created in May 2025. The number is a total, not a weekly gain. The Trendshift rank is the one Trendshift's daily list carried this morning.
Superplane also reports its own usage figures. Both are the company's own. The 1.0 blog post gives a usage figure. Superplane says "52 percent of the pull requests our team merged were automated through Superplane" over one month. That is its own backlog. On its product page, the company says customers reach "30 percent of backlog issues handled autonomously". It also claims a "4x faster issue-to-reviewed-PR cycle". No outside party has verified either figure. The blog post does not say how large the merged pull requests were. It does not say how many were later rolled back.
The product page lists customers: Confluent, Tenderly, Palo Alto Networks, Nutanix, Contentsquare, Delivery Hero, TradingView and Nebius. These are names Superplane writes on its own site. The company has not published a case study with figures from any of them. The 1.0 blog post carries one named testimonial, from Petar Jerinic at Brame.
The company
Superplane was founded in January 2026. It raised a $2.6 million pre-seed round led by Credo Ventures in June 2026. Both dates are from the company's own blog. The team is based in San Francisco.
The release adds two scoring features. The blog post describes both in detail. One is a readiness score. It marks a ticket on clarity, complexity and verifiability before the agent starts. An unclear issue gets flagged for a person. The other is a verification step. It asks for passing checks, screenshots of the real interface and recordings. The agent must say out loud what it could not verify.
A reader can try it two ways. One is the self-hosted version alongside their existing CI. The other is the cloud workspace on superplane.com. The licence, Apache-2.0, lets a team fork the tool and run it on its own servers.
Source
Superplane. Repository on GitHub. Trendshift daily ranking at trendshift.io.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.
