Intelligent test selector for CI based on code change graph

dev tool real project •• multiple requests

Engineers running large test suites manually pick test paths or just run everything and waste minutes per push. The HN 2026 dev-tool wishlist surfaced specific demand for an LLM-assisted tool that proposes the relevant test subset given a diff, plus an estimate of how many iterations are needed to catch flakes. Existing solutions (Launchable, BuildPulse) are enterprise-priced and require pre-existing test history at scale.

builder note

Don't pitch this as ML predictive testing, that name is taken and people associate it with enterprise contracts. Pitch it as 'an MCP server your coding agent already uses' so the test selection happens inline with the agent already touching the code.

landscape (3 existing solutions)

Predictive test selection has been an enterprise category for years. AI coding agents now make a 'just give me a diff and I'll pick the tests' workflow feasible for a single-person OSS project. The gap is a free/cheap, agent-friendly tool that small projects can adopt without a sales call.

Launchable Enterprise-priced predictive test selection. Demo-then-sales-call model. Out of reach for solo devs and small OSS projects that feel the pain most.

BuildPulse Focuses on flake detection rather than diff-aware selection. Different problem shape, requires significant test history to be useful.

Manual jest --findRelatedTests / pytest-testmon Per-language hacks that work on dependency graph or coverage maps. No semantic understanding of 'this diff changed auth so run auth tests AND the integration ones.'

sources (2)

hn https://news.ycombinator.com/item?id=46345827 "LLM tool that analyzes code changes and intelligently proposes relevant test suites" 2025-12-27

other https://blog.zharii.com/blog "Turning repeated rules into deterministic tools like linters, hooks, CI checks" 2026-04-18

ci-cdtestingai-toolstest-selectiondeveloper-tools