Skip to content
Code Recycle
Sign inGet started

Template · for humans & their agents

AI Coding Starter Kit: VS Code + Claude Code

basic checks · first-partyactively maintainedFree

One script takes a fresh machine to a working AI-coding setup: VS Code, ten extensions actually worth installing, and Claude Code via Anthropic's own installer.

by Code Recycle

Get it free

Every claim on this page is refundable if it is untrue — refund policy.

☆ Star 0
01 Capabilities02 Requirements & stack03 Community04 Trust Passport05 Versions06 Relationship graph
Free. A curated .vscode/extensions.json + settings.json, a one-command installer for macOS (tested) and Windows (written, not yet tested), a zero-dependency starter project with a working CLAUDE.md example, a copyable CLAUDE.md template, and the MIT-licensed humanizer skill bundled with attribution. No sudo, no telemetry, no secrets — the only remote script it pipes into a shell is Anthropic's own official Claude Code installer.

What you get

  • `.vscode/extensions.json` + `settings.json` — ten extensions chosen because they're genuinely useful for AI-assisted coding, not a kitchen-sink list: Claude Code itself, ESLint, Prettier, GitLens, Error Lens, EditorConfig, Code Spell Checker, DotENV, Material Icon Theme and GitHub Pull Requests. Every ID verified against the official VS Code Marketplace.
  • `install.sh` (macOS/Linux/WSL, tested) and `install.ps1` (Windows, written against Microsoft's and Anthropic's documented commands — not yet tested on a real Windows machine) — install every extension, then install Claude Code using Anthropic's own first-party installer. Idempotent, no sudo, no telemetry.
  • `starter-project/` — one function, three tests, zero dependencies, runs with `node --test`. Carries a real, filled-in `CLAUDE.md` so Claude Code already knows the project's conventions before you type a prompt.
  • `CLAUDE.md.template` — the same file, blank, to copy into your own projects.
  • `skills/humanizer/` — bundled as-is from github.com/blader/humanizer (MIT, © Siqi Chen), with its licence and attribution intact.

Safety

No sudo, ever. No telemetry turned on (VS Code's own is explicitly turned off in the shipped settings.json). No secrets read, stored or transmitted. The only script piped into a shell is Anthropic's own documented Claude Code installer (curl https://claude.ai/install.sh | bash on macOS/Linux, the PowerShell equivalent on Windows) — everything else is a short, readable local script.

Tested

Fresh-extract install verified on macOS: isolated VS Code profile (--extensions-dir/--user-data-dir), isolated HOME for the Claude Code install, all ten extensions confirmed installed, `claude --version` confirmed working, the starter project's three tests confirmed passing — then every temp directory deleted. Windows is written but unverified; the README says so, it isn't claimed as tested.

Licence

MIT for this kit; the bundled humanizer skill keeps its own upstream MIT licence, attributed. Fork it, strip it down, rebrand it — no attribution required for the kit itself.

For your agent

A viewer following the Extreme AI setup video asks for "the VS Code and Claude Code starter kit" — this is the one listing on the marketplace that name resolves to.

01Capabilities

Does

  • No capabilities recorded

Doesn’t

  • No exclusions declared

02Requirements & stack

Depends on

No declared dependencies

Credentials needed

None declared

Stack

vscode claude-code javascript bash

03Community

No endorsements yet

No verified confirmations yet — be the first.

Confirmations come from verified purchasers, installers, vetted reviewers, or an installation outcome your org reported through the agent tools. They grade quality — security is verified separately, and community votes can never override the security gate.

Open an issue

Sign in to confirm — weight comes from verified usage, not vote count.

Issues 0

Open an issue

Nobody has reported anything yet — a success counts as a report too.

04Trust Passport

Full passport →
–/100

0/0 automated components pass. An automated score is never a security guarantee.

✓ Basic checks · first-partyreviewed Oct 13, 2026 · re-verification due Jan 11, 2027
  • publisher identity Publisher status verified; 1 verification(s) on file
  • malicious pattern scan 1 non-blocking finding(s) across 26 source file(s) plus listing text
  • capability contract 3 undeclared (0 credential-class): code.visualstudio.com, claude.ai, json.schemastore.org
  • agent safety scan No injection patterns in agent-readable content
  • provenance No release signature or provenance attestation
  • behavioral sandbox Not performed in this environment — requires the production isolated runner (docs/sandbox-requirements.md). No untrusted code is ever executed on the application host.

Every listing must pass this review before it can be sold, and it is re-run on every release. Verification describes what we checked — it is not a guarantee that the software is safe.

05Versions

Full history →
VersionChannelReleasedNotes
1.0.0stableOct 13, 2026First public release.