@alumbwe/anvil@1.0.69
A Terminal User Interface (TUI) package built with OpenTUI and React.
AI Security Review
scanned 3h ago · by lpm-firewall-aiThe package uses conversation text in ad-targeting requests that can run with ads enabled by default. Because this goes to the Anvil service for a feature the package provides, the evidence supports a warning for insufficiently disclosed content collection rather than a malicious exfiltration verdict.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
Normal CLI use when ads are enabled; Anvil enables them by default.
Impact
Conversation content may be disclosed to the Anvil service for ad targeting; the source does not establish that this is hidden collection or that it is sent to an unrelated host.
Mechanism
The ad hook extracts user and assistant text from message history and includes it in a POST request with the session ID and device information.
AI rationale
The package uses conversation text in ad-targeting requests that can run with ads enabled by default. Because this goes to the Anvil service for a feature the package provides, the evidence supports a warning for insufficiently disclosed content collection rather than a malicious exfiltration verdict.
Evidence
src/hooks/use-gravity-ad.tssrc/commands/ads.ts
Network endpoints1
${WEBSITE_URL}/api/v1/ads
Decision evidence
public snapshotAI called this Suspicious at 82.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The CLI posts ad requests to `${WEBSITE_URL}/api/v1/ads` with conversation messages and a session ID.
- The ad request builder includes user and assistant text, excluding messages tagged `INSTRUCTIONS_PROMPT`.
- Ads are enabled by default for Anvil, so this transmission can occur during normal CLI use.
Evidence against
- The code describes the ads as contextual, suggesting conversation context is used to provide the ad feature.
- The ad request goes to the configured Anvil website endpoint, not an unrelated collector.
- The flagged Bun installer runs only when the CLI starts and Bun is missing; it invokes npm to install Bun.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetworkShellWebSocket
HighEntropyStringsTelemetryUrlStrings
Source & flagged code
9 flaggedbin/anvil.jsView file
17L18: import { spawnSync } from 'child_process'
L19: import fs from 'fs'
High
5* Runs on plain Node.js (Bun is NOT required upfront) and works from cmd,
L6: * PowerShell and POSIX shells: npm generates a shim for each shell that all
L7: * execute this file with `node`.
High
117if (process.versions.bun) {
L118: await import(pathToFileURL(ENTRY).href)
L119: } else {
Medium
Dynamic Require
Package source references dynamic require/import behavior.
bin/anvil.jsView on unpkg · L117vendor/@anvil/sdk/dist/index.cjsView file
Findings
1 Critical4 High6 Medium8 Low
CriticalTrojan Source Unicodevendor/@anvil/sdk/dist/index.cjs
HighChild Processbin/anvil.js
HighShellbin/anvil.js
HighObfuscated Payload Loadervendor/@anvil/sdk/dist/index.cjs
HighRemote Agent Bridgevendor/@anvil/sdk/dist/index.cjs
MediumDynamic Requirebin/anvil.js
MediumNetwork
MediumEnvironment Vars
MediumShips Native Binaryvendor/@anvil/sdk/dist/vendor/ripgrep/x64-darwin/rg
MediumShips Wasm Modulebin/tree-sitter.wasm
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
Affected versions and remediation
This report applies to @alumbwe/anvil@1.0.69.
See all versions of @alumbwe/anvil for other recorded verdicts.
- Review the evidence and your use of @alumbwe/anvil@1.0.69 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.