Agent skill · Audited from source

Amazon

skill · vellum-ai/vellum-assistantMIT licenceAudited 14 Oct 2026 · source 19bc86343b63

★ 1,425pushed 1 day ago
✓SAFEGrade B · 89/100 · 14 Oct 2026 · 19bc863
1,425repo stars · whole vellum-ai/vellum-assistant

Is it safe? Yes — SAFE, grade B (89/100). We read this skill's source at commit 19bc86343b63 and found no filesystem, network or shell access and no hard-coded secrets. What it does: Shop on Amazon and Amazon Fresh through your browser

Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
Cite

Cite this page

ModelsAtlas. “Amazon (vellum-ai/vellum-assistant)”. Retrieved 14 Oct 2026. https://modelsatlas.com/skills/vellum-ai/amazon

Trust audit

Audit 0.4.1 · source read 14 Oct 2026

Trust score89/100✓SAFE B
0 critical0 high0 medium5 low0 info

SAFE means nothing in the source contradicts what the skill says it does. Grade B is shared by 13,985 of 14,816 audited skills.

What the source does

  • FilesystemNone observedNo reads or writes in the source
  • NetworkNone observedNo outbound calls in the source
  • ShellNone observedNo commands run in the source
  • DependenciesPinnedNothing floats
  • SecretsNone foundNo keys in source

Five audit layers

LayerWhat we checkResultFindings
L0 ProvenanceRepo owner, licence, archive status, activityPASS5
L1 Static sourceSecrets, path traversal, supply chain, hidden codePASS0
L2 InstructionsPrompt injection, authority framing, hidden directivesPASS0
L3 InventoryThe files and declarations match the sourcePASS0
L4 SandboxRun in isolation and watch behaviourSKIPPED why0

Gates applied: no behavioural pass.

Findings (5)

Audit 0.4.1 · source sha 19bc86343b63 · 14 Oct 2026Report an issue or request a re-scan

Provenance

Who published it, and what the verdict is pinned to

  • Publishervellum-ai GitHub organizationOwner of the repository this skill is read from
  • Source foldervellum-ai/vellum-assistant/skills/amazonThe exact folder we audited
  • LicenceMITAs the repository states it
  • Commit pinned for this verdict19bc86343b63A new audit runs when the source changes
  • Repository statusNot archivedLast push 1 day ago
  • Sandboxed runNot runSkills are read, not executed; layer L4 is skipped

What it tells the agent

Full source on GitHub ↗

Its description: Shop on Amazon and Amazon Fresh through your browser

---
name: amazon
description: Shop on Amazon and Amazon Fresh through your browser
compatibility: "Designed for Vellum personal assistants"
metadata:
  icon: assets/icon.svg
  emoji: "🛒"
  vellum:
    category: "commerce"
    display-name: "Amazon"
    includes: ["vellum-browser-use"]
---

Use browser automation for all Amazon actions. All browser operations are executed through the `assistant browser` CLI, invoked via `host_bash`. Use helper scripts with `host_bash` to normalize extraction results and decide the next step.

## Required tools

- `host_bash` for `assistant browser` CLI commands and deterministic helper scripts under `scripts/`.

## Hard constraints

- Do not call `assistant browser chrome relay`.
- Do not use legacy relay-backed scripts.
- Always require explicit user confirmation before final order submission.

## Step graph (state machine)

### Step 1: Classify workflow state

Run this early in each turn when intent is unclear:

```bash
bun {baseDir}/scripts/amazon-intent.ts --request "<latest user request>" --checkout-reviewed <true|false> --has-cart-items <true|false>
```

Use the returned `step` to route to one of: `search`, `variant_select`, `cart_review`, `checkout_review`, `fresh_slot`, `place_order`.

### Step 2: Product discovery (`search`)

1. Navigate to search results page:

```bash
assistant browser --session amazon navigate --url "https://www.amazon.com/s?k=<urlencoded query>"
```

2. Capture current state:

```bash
assistant browser --session amazon --json snapshot
assistant browser --session amazon --json extract --include-links
```

3. Parse candidates deterministically:

```bash
bun {baseDir}/scripts/amazon-parse-search.ts --query "<query>" --input-json '<json payload with extracted text/links>'
```

4. Present top options with title, price, ASIN (if present), Prime/Fresh hints.

### Step 3: Product detail + variant resolution (`variant_select`)

1. Open product result.
2. Re-snapshot + re-extract.
3. Parse product details:

```bash
bun {baseDir}/scripts/amazon-parse-product.ts --input-json '<json payload with extracted text/links>'
```

4. If variation hints are present, resolve user choice before add-to-cart.

### Step 4: Add to cart + verify (`cart_review`)

1. Click Add to Cart on product page.
2. Navigate to cart page and extract:

```bash
assistant browser --session amazon navigate --url "https://www.amazon.com/gp/cart/view.html"
assistant browser --session amazon --json snapshot
assistant browser --session amazon --json extract --include-links
```

3. Parse cart summary:

```bash
bun {baseDir}/scripts/amazon-parse-cart.ts --input-json '<json payload with extracted text>'
```

4. Show parsed line items and totals. Ask user to confirm cart contents.

### Step 5: Fresh slot validation (`fresh_slot`)

For Amazon Fresh flows, explicitly verify slot selection in UI before checkout:

1. Navigate to Fresh delivery slot surface if needed.
2. Snapshot + extract delivery slot details.
3. Confirm selected slot text is visible before proceeding.

If slot cannot be verified after retries, stop and ask user to choose slot manually.

### Step 6: Checkout sanity (`checkout_review`)

1. Navigate to checkout review page.
2. Snapshot, extract, and capture a full-page screenshot:

```bash
assistant browser --session amazon --json snapshot
assistant browser --session amazon --json extract
assistant browser --session amazon screenshot --full-page --output /tmp/amazon-checkout.jpg
```

3. Validate readiness:

```bash
bun {baseDir}/scripts/amazon-checkout-sanity.ts --cart-confirmed true --input-json '<json payload with extracted text>'
```

4. Report missing markers (shipping/payment/total/submit action) before any submission.

### Step 7: Final submit gate (`place_order`)

Immediately before clicking final submit button:

1. Ask for explicit final confirmation in plain language.
2. If user confirms, click final submit action (`Place your order`, `Buy now`, or equivalent).
3. Take post-submit snapshot/screenshot and report confirmation details.

## Retry and fallback policy

- Retry budget: 3 attempts per step that mutates page state.
- After each mutation, run a fresh `assistant browser --session amazon --json snapshot` before the next click/type.
- If a step fails 3 times, stop and ask user to complete that step manually, then resume.

## Example helper payload shape

```json
{
  "phase": "search",
  "context": { "checkoutReviewed": false, "hasCartItems": false },
  "extracted": {
    "text": "...",
    "links": ["https://www.amazon.com/dp/B08XGDN3TZ"]
  },
  "userIntent": "order aa batteries"
}
```

## Safety rules

- Always show price/totals before confirmation.
- Never infer final consent from prior messages; ask again right before submission.
- If CAPTCHA or anti-bot challenge appears, ask user to solve it and continue after refresh.

Excerpt, first 4,847 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.

Install

Pinned to the audited commit 19bc863

Download folder ↗
git clone https://github.com/vellum-ai/vellum-assistant
git -C vellum-assistant checkout 19bc86343b63

The author documents no install command, so this uses the repository directly. Our verdict covers commit 19bc86343b63 only.

Works in

The author has not said which agents this skill supports, and we have not tested it in any.

Audit history

Every audit, every source change

  1. ✓SAFE B 89 / 100

    first audit · source 19bc86343b63

README badge

For maintainers · shows the latest audit

ModelsAtlas: SAFE BMODELSATLASSAFE · B
[![ModelsAtlas: SAFE B](https://modelsatlas.com/skills/vellum-ai/amazon/badge.svg)](https://modelsatlas.com/skills/vellum-ai/amazon/)

The badge shows our verdict and grade for this skill, and links back to this page. It changes when the audit does.

Add this badge to your README

[![ModelsAtlas: SAFE B](https://modelsatlas.com/skills/vellum-ai/amazon/badge.svg)](https://modelsatlas.com/skills/vellum-ai/amazon/)

Questions about Amazon

What does the Amazon skill do?
An AI Assistant that’s easy to setup, does your work 24/7, knows your preferences and gets better over time.
Is Amazon safe to install?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.
What can Amazon access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
How current is this page?
The grade is for one exact copy of the source (19bc86343b63), read on 2026-10-14. The repository is watched, and a new audit runs when it changes — this is the first audit.

Sources

Spotted something wrong? Report a correction · Methodology

Keep going in your AI

Opens with this page as the source. We add nothing else to the prompt.