Each colour maps a published value to the exact source passage used to support it. Only bounded excerpts are public; administrators can inspect the complete captured source.
Title
Spotify adds background coding agents to its Fleet Management platform, merging over 1,500 AI-generated pull requests
…gration still requires many human hands. What if we could close that final gap? This post explores how we have evolved our Fleet Management platform with background coding agents and what more than 1,500 merged AI-generated pull requests have taught us about the future of large-scale software maintenance. Automating toil using Fle…
Description
Spotify replaced deterministic migration scripts in its Fleet Management system with coding agents driven by natural-language prompts, built around a small internal CLI that supports multiple agents and LLMs, local MCP tooling, LLM-as-judge diff evaluation and MLflow tracing. The source reports more than 1,500 merged agent-generated pull requests, a 60-90% time saving on migrations compared with hand-written changes, and agent tasks that can be started from Slack and GitHub Enterprise.
…s that needed the most help: the declaration of the code transformation itself. We replaced deterministic migration scripts with an agent that takes instructions from a prompt. All the surrounding Fleet Management infrastructure — targeting repositories, o…
…off-the-shelf coding agent as it is, we decided to build a small internal CLI. This CLI can delegate executing a prompt to an agent, run custom formatting and linting tasks using local Model Context Protocol (MCP), evaluate a diff using LLMs as a judge, upload logs to Google Cloud Platform (GCP), and capture traces in MLflow. Crucially, having that CLI allows us to seamlessly switch between different age…
…tooling alongside early adopters who applied it to their in-flight migrations. To date, our agents have generated more than 1,500 pull requests that teams across Spotify have merged into our production codebase. And not trivial changes, either — we’re now starting to tackle changes such as:…
…g parameters in YAML or JSON files and still adhering to schemas and formatting For these migrations we’ve seen a total time saving of 60–90% compared to writing the code by hand the good old-fashioned way. The ROI for an automated change also continues to increase as we scale it out t…
…multi-agent architecture for planning, generating, and reviewing pull requests. By exposing our background coding agent via MCP, Spotifiers can now kick off coding agent tasks from both Slack and GitHub Enterprise. They first talk to an interactive agent that helps to gather information about…
Company
Spotify
1,500+ PRs Later: Spotify’s Journey with Our Background Coding Agent (Honk, Part 1) | Spotify Engineering 1,500+ PRs Later: Spotify’s Journey with Our Background Coding Agent (Honk, Par…
Industry
Technology & Software
…0+ PRs Later: Spotify’s Journey with Our Background Coding Agent (Honk, Part 1) This is part 1 in our series about Spotify's journey with background coding agents (internal codename: “Honk”) and the future of large-scale software maintenance. See also part 2 , part 3 , and part 4 . For years, developer productivity has i…
Problem
Spotify's Fleet Management system automated much of the toil of maintaining its codebase, but any moderately complex migration still required many human hands, because defining source code transformations programmatically requires specialised expertise (one Maven dependency updater script grew to over 20,000 lines of code).
…hat the code follows best practices demands a surprising amount of manual work. At Spotify, our Fleet Management system automated much of that toil, yet any moderately complex migration still requires many human hands. What if we could close that final gap? This post explores how we have evolved o…
…ble tasks, making complex code changes is a challenge we’ve never fully solved. Defining source code transformations programmatically, by manipulating a program’s abstract syntax tree (AST) or using regular expressions, requires a high degree of specialized expertise. One example is our automated Maven dependency updater. While its core function…
…identifying pom.xml files and updating Java dependencies — is straightforward, handling all corner cases has led to the transformation script growing to over 20,000 lines of code. Because of this complexity, most of the automated changes we could implement w…
Solution
Spotify replaced deterministic migration scripts with an agent that takes instructions from a prompt, leaving the surrounding Fleet Management infrastructure (targeting repositories, opening pull requests, reviews and merging) unchanged. It built a small internal CLI that delegates a prompt to an agent, runs formatting and linting through local MCP, evaluates diffs with LLMs as a judge, and captures traces in MLflow, so it can switch between agents and LLMs. The agent is also exposed via MCP so tasks can be started from Slack and GitHub Enterprise.
…s that needed the most help: the declaration of the code transformation itself. We replaced deterministic migration scripts with an agent that takes instructions from a prompt. All the surrounding Fleet Management infrastructure — targeting repositories, o…
…ministic migration scripts with an agent that takes instructions from a prompt. All the surrounding Fleet Management infrastructure — targeting repositories, opening pull requests, getting reviews, and merging into production — remains exactly the same. Figure 4: Configuring a Fleetshift with a prompt. Instead of adopting an off-th…
…udge, upload logs to Google Cloud Platform (GCP), and capture traces in MLflow. Crucially, having that CLI allows us to seamlessly switch between different agents and LLMs. In the fast-moving environment that is GenAI, being flexible and pluggable this…
…multi-agent architecture for planning, generating, and reviewing pull requests. By exposing our background coding agent via MCP, Spotifiers can now kick off coding agent tasks from both Slack and GitHub Enterprise. They first talk to an interactive agent that helps to gather information about…
Business value
Spotify reports that its agents have generated more than 1,500 pull requests that teams have merged into the production codebase, and a total time saving of 60–90% on these migrations compared with writing the code by hand. The source also notes open challenges around performance, predictability, safety and cost.
…tooling alongside early adopters who applied it to their in-flight migrations. To date, our agents have generated more than 1,500 pull requests that teams across Spotify have merged into our production codebase. And not trivial changes, either — we’re now starting to tackle changes such as:…
…g parameters in YAML or JSON files and still adhering to schemas and formatting For these migrations we’ve seen a total time saving of 60–90% compared to writing the code by hand the good old-fashioned way. The ROI for an automated change also continues to increase as we scale it out t…
…ictable. This creates a need for new validation and quality control mechanisms. Beyond performance and predictability, we also have to consider safety and cost. We need robust guardrails and sandboxing to ensure agents operate as intended,…
AI capabilities
Agentic AI, Generative AI
…tooling alongside early adopters who applied it to their in-flight migrations. To date, our agents have generated more than 1,500 pull requests that teams across Spotify have merged into our production codebase. And not trivial changes, either — we’re now starting to tackle changes such as:…
Headline outcome
…g parameters in YAML or JSON files and still adhering to schemas and formatting For these migrations we’ve seen a total time saving of 60–90% compared to writing the code by hand the good old-fashioned way. The ROI for an automated change also continues to increase as we scale it out t…
Use case type
Software engineering assistance
…ger the same agent from your IDE or chat, give it a task, and then go to lunch? Over time, it became clear that there is a need for a background coding agent not only for migrations, but also for ad hoc tasks. Figure 6: Current multi-agent architecture for planning, generating, and review…
