AI coding tools · VS Code trust review

VS Code AI extension permissions checklist

Use this VS Code AI extension permissions checklist before an AI coding tool sees a private repository, reads workspace context, opens a terminal, or routes prompts through a provider path your team has not reviewed.

Compare with extension security Install LegacyDoc AI

Quick answer

A VS Code AI extension permissions checklist should answer six questions before install: what workspace files can the tool use, where code or prompts go, whether it can run commands, what secrets are nearby, which VS Code profile should contain it, and what evidence proves the review happened.

Official VS Code doc

Workspace Trust

VS Code explains Restricted Mode and trust prompts for unfamiliar workspaces.

Extension API

Workspace Trust guide

Extension authors can declare and gate behavior for untrusted workspaces.

Extension manifest

Capabilities and package.json

The manifest records extension metadata, contributions, and capabilities.

Why this page exists

AI extensions need a permissions review, not just a publisher glance

A normal extension review often starts with publisher trust, marketplace history, repository links, and release cadence. That still matters. But AI coding extensions add a second layer: they can collect context, summarize files, generate patches, ask a model to reason about private code, and sometimes suggest commands that run inside your local environment.

The right question is not simply "is this extension popular?" The right question is whether the extension belongs inside this exact workspace, with these secrets nearby, using this provider route, under this team policy. That is why a VS Code AI extension permissions checklist should be specific enough for a developer to use before the first install.

VS Code has security controls such as Workspace Trust, Restricted Mode, publisher trust signals, and extension metadata. Those controls are useful, but they do not replace a team decision about AI context flow. If the extension will touch a client repository, production-adjacent codebase, or paid product, treat the review as part of your software supply chain.

Workspace files and repository scope

Decide which folders the AI extension can see, whether the extension reads entire files or selected snippets, and whether client code, private product logic, generated code, or production configuration will be open while it runs.

AI provider and network path

Confirm whether prompts go directly to your chosen model provider, through a vendor proxy, through a company gateway, or to a hosted service you cannot inspect. BYOK is useful only when the route is clear.

Commands, terminals, tasks, and changes

Check whether the extension can run commands, open terminals, edit files, generate patches, trigger tests, or modify project settings. These are not just convenience features; they change the risk model.

Secrets and local credentials

Review what is visible in the workspace and developer machine: .env files, shell history, npm tokens, GitHub tokens, cloud profiles, local database URLs, SSH keys, and AI provider keys.

Profiles, trust, and team policy

Use separate VS Code profiles for experimental AI tools, client work, and production-adjacent code. Decide whether the extension belongs in a team allowlist before it appears on every workstation.

Audit evidence

Record the extension ID, publisher, version, install date, approved repositories, provider route, privacy boundary, and reviewer. A short note is better than an approval that no one can reconstruct.

Review workflow

Run the VS Code AI extension permissions checklist in this order

1. Start with the workspace, not the marketplace page

First decide which repository you are protecting. A hobby repo, a client repo, a closed-source product, and a production incident workspace do not deserve the same default policy. If the workspace contains private business logic, customer data, API keys, cloud credentials, database URLs, or unreleased product code, use a stricter profile and a smaller trial.

2. Map the AI context path

Identify whether the extension reads selected text, open files, entire folders, search results, dependency graphs, terminal output, or generated context packs. Then identify where that context goes. The provider path matters: direct BYOK to your model provider is different from an opaque proxy, shared service, or unknown logging pipeline.

3. Separate read features from action features

Reading a file is one risk. Editing many files, creating commits, running shell commands, modifying settings, or starting local tasks is another. A useful VS Code AI extension permissions checklist separates "can inspect" from "can change" and asks whether a human must approve commands, patches, and generated migrations.

4. Check secrets proximity before opening the repo

Many teams discover the secret problem too late. They test an AI extension in the same workspace where .env files, deployment configs, local database URLs, GitHub tokens, npm tokens, cloud profiles, and shell history are visible. Move secrets out of project files where possible and keep experimental tools away from high-risk workspaces.

5. Record the approval so future you can audit it

A one-line approval is not enough. Record the extension ID, publisher, version, date, approved repositories, provider path, privacy note, and reviewer. If the extension is later removed, compromised, or re-scoped, this note tells you which repositories and credentials need a second look.

Risk matrix

Permission questions that matter for AI coding extensions

Review area	Why it matters	Practical control
Workspace read access	Private source files, customer logic, generated code hotspots	Limit the workspace, use profiles, and avoid opening sensitive repos during trial runs.
Provider routing	Code or prompts may leave the machine through an unclear path	Prefer direct provider connections, documented BYOK, or a known company gateway.
Terminal and task execution	Generated commands can touch files, services, databases, or credentials	Require manual review before running commands; keep high-risk terminals out of experiments.
Secrets proximity	Tokens may sit beside the tool in .env files, configs, history, or cloud profiles	Move secrets out of project files and rotate credentials after suspicious extension behavior.
Auto-edits and refactors	Large patches can hide unrelated changes or downgrade tests	Review diffs, run tests, and keep AI-generated changes small enough to understand.
Telemetry and logs	Prompts, filenames, errors, or snippets may appear in logs	Check privacy docs, disable unnecessary telemetry where possible, and keep sensitive examples out of prompts.

Copyable approval note

Template for approving an AI extension in VS Code

Extension:
Publisher:
Version:
Marketplace URL:
Approved workspace type:
Blocked workspace type:
Provider path:
BYOK or vendor key:
Can read workspace files:
Can run commands or terminals:
Can edit files:
Secrets policy:
VS Code profile:
Reviewer:
Review date:
Rollback note:

This note makes the VS Code AI extension permissions checklist operational. It is short enough for a solo founder, but structured enough for a small team to revisit when the extension changes behavior or a security warning appears.

How RomantiCode fits

Prefer local-first documentation tools with a clear provider boundary

LegacyDoc AI is built for teams that need codebase maps, documentation, and cleanup context from inside VS Code. The trust boundary is deliberately simple: the extension runs in VS Code, you bring your own AI provider key, and RomantiCode does not store your code. That does not remove your responsibility to review the provider you choose, but it makes the extension boundary easier to explain during a permissions review.

If you are preparing an AI-built app for cleanup, use this VS Code AI extension permissions checklist together with the broader VS Code extension security checklist, the RomantiCode privacy policy, and the AI code audit report workflow. The result is a clearer story for developers, clients, and future reviewers.

MCP review

MCP security scanner

Review MCP server tools, files, secrets, network egress, and agent trust boundaries.

General security

VS Code extension security checklist

Review publisher trust, updates, secrets, and incident response.

Audit workflow

AI code audit report

Create a cleanup-ready context pack before handoff or review.

Launch review

AI app launch audit

Scope production readiness, cleanup risk, and release blockers.

FAQ

VS Code AI extension permissions checklist FAQ

What is a VS Code AI extension permissions checklist? +

It is a practical review list for AI coding extensions. It helps a developer or team check workspace access, provider routing, terminal behavior, secrets exposure, profiles, and approval evidence before the extension is used on private code.

Does VS Code show one permission prompt for every extension capability? +

No single prompt answers the whole review question for a team. VS Code has safety features such as Workspace Trust and extension trust signals, but teams still need to review the extension's behavior, data path, and local risk boundary.

How is this different from a VS Code extension security checklist? +

A general extension security checklist covers publisher trust, malware risk, updates, and incident response. This page focuses on AI-specific permission questions: code context, model provider path, generated changes, terminals, secrets, and team approval.

Should I use a separate VS Code profile for AI extensions? +

For sensitive work, yes. Separate profiles make it easier to keep experimental AI tools away from client repos, production-adjacent code, and workspaces that contain credentials or regulated data.

How does LegacyDoc AI handle code and API keys? +

LegacyDoc AI runs inside VS Code and uses the AI provider you configure. Your code is sent directly to that provider, not to RomantiCode servers. You bring your own API key.

Next step

Review AI tools before they touch the wrong workspace.

Use this VS Code AI extension permissions checklist, then generate a codebase map and audit report for the repos that are safe to review with AI assistance.

Install LegacyDoc AI Review privacy boundary

SEO audit support

VS Code AI extension permissions checklist decision checklist

This section keeps VS Code AI extension permissions checklist focused on one search intent. A reader comparing options for VS Code AI extension permissions checklist should quickly see the task, the evidence, the handoff value, and the next action without leaving the page.

VS Code AI extension permissions checklist workflow screenshot for RomantiCode SEO audit — RomantiCode uses real VS Code context to support VS Code AI extension permissions checklist decisions before cleanup, audit, or handoff.

VS Code AI extension permissions checklist checkpoint 1

Use VS Code AI extension permissions checklist as the page promise, then verify that VS Code AI extension permissions checklist is supported by the headline, the example, the internal links, the call to action, and the reader's next step. A strong VS Code AI extension permissions checklist page should explain who needs VS Code AI extension permissions checklist, what evidence is required before acting, and how RomantiCode reduces uncertainty for founders, developers, cleanup specialists, and AI coding agents. Keep the VS Code AI extension permissions checklist checklist tied to a real workflow: inspect the codebase, map risky files, prepare context, compare options, and decide whether to audit, refactor, hire help, or continue with an AI assistant.

VS Code AI extension permissions checklist

Workspace Trust

Workspace Trust guide

Capabilities and package.json

AI extensions need a permissions review, not just a publisher glance

Workspace files and repository scope

AI provider and network path

Commands, terminals, tasks, and changes

Secrets and local credentials

Profiles, trust, and team policy

Audit evidence

Run the VS Code AI extension permissions checklist in this order

1. Start with the workspace, not the marketplace page

2. Map the AI context path

3. Separate read features from action features

4. Check secrets proximity before opening the repo

5. Record the approval so future you can audit it

Permission questions that matter for AI coding extensions

Template for approving an AI extension in VS Code

Prefer local-first documentation tools with a clear provider boundary

MCP security scanner

VS Code extension security checklist

AI code audit report

AI app launch audit

VS Code AI extension permissions checklist FAQ

Review AI tools before they touch the wrong workspace.

VS Code AI extension permissions checklist decision checklist

VS Code AI extension permissions checklist checkpoint 1

VS Code AI extension permissions checklist checkpoint 2

VS Code AI extension permissions checklist checkpoint 3

VS Code AI extension permissions checklist checkpoint 4

VS Code AI extension permissions checklist checkpoint 5

VS Code AI extension permissions checklist checkpoint 6

VS Code AI extension permissions checklist checkpoint 7

VS Code AI extension permissions checklist checkpoint 8

VS Code AI extension permissions checklist checkpoint 9

VS Code AI extension permissions checklist checkpoint 10

VS Code AI extension permissions checklist checkpoint 11

VS Code AI extension permissions checklist checkpoint 12

VS Code AI extension permissions checklist checkpoint 13

VS Code AI extension permissions checklist checkpoint 14

VS Code AI extension permissions checklist checkpoint 15

VS Code AI extension permissions checklist checkpoint 16

VS Code AI extension permissions checklist checkpoint 17

VS Code AI extension permissions checklist checkpoint 18

VS Code AI extension permissions checklist checkpoint 19

VS Code AI extension permissions checklist checkpoint 20

VS Code AI extension permissions checklist checkpoint 21

VS Code AI extension permissions checklist checkpoint 22

VS Code AI extension permissions checklist checkpoint 23

VS Code AI extension permissions checklist checkpoint 24

VS Code AI extension permissions checklist checkpoint 25

VS Code AI extension permissions checklist checkpoint 26

VS Code AI extension permissions checklist checkpoint 27

VS Code AI extension permissions checklist checkpoint 28

VS Code AI extension permissions checklist checkpoint 29

VS Code AI extension permissions checklist checkpoint 30

VS Code AI extension permissions checklist checkpoint 31

VS Code AI extension permissions checklist checkpoint 32

VS Code AI extension permissions checklist checkpoint 33

VS Code AI extension permissions checklist checkpoint 34

VS Code AI extension permissions checklist checkpoint 35

VS Code AI extension permissions checklist checkpoint 36

VS Code AI extension permissions checklist checkpoint 37

VS Code AI extension permissions checklist checkpoint 38

VS Code AI extension permissions checklist checkpoint 39

VS Code AI extension permissions checklist checkpoint 40

VS Code AI extension permissions checklist checkpoint 41

VS Code AI extension permissions checklist checkpoint 42

VS Code AI extension permissions checklist checkpoint 43

VS Code AI extension permissions checklist checkpoint 44

VS Code AI extension permissions checklist checkpoint 45

VS Code AI extension permissions checklist checkpoint 46

VS Code AI extension permissions checklist checkpoint 47

VS Code AI extension permissions checklist checkpoint 48

VS Code AI extension permissions checklist checkpoint 49

VS Code AI extension permissions checklist checkpoint 50

VS Code AI extension permissions checklist checkpoint 51

VS Code AI extension permissions checklist checkpoint 52