buildNext

Working notes and architecture documentation for VS Code’s new esbuild-based build system, which turns TypeScript into JavaScript and bundled files.

In plain words
What is it for?
Use it when changing transpiling, bundling, language-file plugins, source maps, resource copying, or watch tasks in build/next.
Why use it?
It gives coding agents the project context and commands needed to change and check the new build pipeline without relying on the main VS Code build task.

Instructions file for GitHub Copilot

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/microsoft/vscode/buildnext
Clone the repo
git clone --depth 1 https://github.com/microsoft/vscode

Made for: GitHub Copilot.

Per session 6,785 This file is loaded in full into every session.
When invoked 6,785 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.06785 $0.06785
Opus 5 $0.03393 $0.03393
Sonnet 5 $0.01357 $0.01357
Haiku 4.5 $0.00679 $0.00679

Measured yesterday against content hash 15fe3cdbaefe, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

buildNext scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.github/instructions/buildNext.instructions.md · 347 lines

How it starts

The opening of the file, as written. The whole thing — 347 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Working Notes: New esbuild-based Build System

These notes are for AI agents to help with context in new or summarized sessions.

Important: Validating Changes

The VS Code - Build task is NOT needed to validate changes in the build/ folder!

Build scripts in build/ are TypeScript files that run directly with node (e.g., node build/next/index.ts). They are not compiled by the main VS Code build.

To test changes:

# Test transpile
node build/next/index.ts transpile --out out-test

# Test bundle (server-web target to test the auth fix)
node build/next/index.ts bundle --nls --target server-web --out out-vscode-reh-web-test

# Verify product config was injected
grep -l "serverLicense" out-vscode-reh-web-test/vs/code/browser/workbench/workbench.js

Architecture Overview

Files

  • index.ts - Main build orchestrator
    • build-fast command: Persistent non-watch incremental development build
    • transpile command: Fast TS → JS using esbuild.transform()
    • bundle command: TS → bundled JS using esbuild.build()
  • build-fast.ts - Git change discovery, persistent state, lane planning, and orchestration
  • transpile.ts - Shared full/watch/incremental transpile and copy operations
  • nls-plugin.ts - NLS (localization) esbuild plugin
  • private-to-property.ts - Native private to property transformation

Fast Non-Watch Incremental Builds

npm run build-fast stores its last successful input state in .build/build-fast/state.json. It uses scoped Git deltas plus content hashes for currently dirty/untracked inputs, so a repeated invocation can skip every build lane without running a watcher or scanning all source files.

  • Client changes are transpiled/copied/deleted at file granularity.
  • Built-in extension/media and Copilot builds run only when their inputs change.
  • Missing, incompatible, or invalid state falls back to a full build.
  • npm run build-fast -- --force forces all lanes to rebuild and refreshes state.
  • State is invalidated before outputs change and published only after every selected lane succeeds. If inputs change during the build, the pre-build snapshot is saved so late changes are rebuilt on the next run.

Read the full file on GitHub · 347 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. yesterday First seen · 347 lines · 6,785 tokens per session scan A 15fe3cdbaefe

Subscribe to this mod's changes

buildNext is an instructions file published in the GitHub repository microsoft/vscode (190,061 stars, last pushed yesterday), licensed MIT. It adds 6,785 tokens to every session, about $0.0339 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.