@ubnt/cawcut@0.6.0

CawCut developer CLI — auth, generate, app run, task status

AI Security Review

scanned 10d ago · by lpm-firewall-ai

The global postinstall can automatically add package-owned skills to agent skill directories, creating a real extension lifecycle risk. The behavior is scoped and has skip conditions, with no inspected evidence of a concrete attack; downgrade to a warning.

Static reason

High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review

Trigger

Installing the package globally, unless the lifecycle skip conditions apply.

Impact

The installed skills can affect instructions available to those agents.

Mechanism

The postinstall routine invokes a skill installer for named package skills, targeting the default agents and using the package’s GitHub skills repository by default.

AI rationale

Evidence

package.jsondist/postinstall.jsdist/lib/is-global-install.jsdist/lib/skills-install.jsdist/lib/skills-agents.js~/.cursor/skills~/.claude/skills~/.codex/skills~/.agents/skills

Network endpoints1

github.com

Decision evidence

public snapshot

AI called this Suspicious at 89.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

The manifest runs a postinstall script that enters the package’s setup routine.
On global installs, the routine installs named package skills for the default editor agents without an interactive confirmation.
The installer targets global skill directories and defaults to the package’s GitHub skills repository.
This is a first-party agent skill setup that can change the instructions available to installed agents; the source does not establish malicious payload behavior.

Evidence against

The setup skips CI and supports an explicit skip variable, and it does not run for local installs.
The installation is scoped to named skills and the listed agents, rather than broad agent configuration changes.

Behavioral surface

Source

ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsUrlStrings

Manifest

Source & flagged code

6 flagged

package.jsonView file

Lines 1-40json

2 "name": "@ubnt/cawcut",

3 "version": "0.6.0",

4 "description": "CawCut developer CLI \u2014 auth, generate, app run, task status",

5 "type": "module",

6 "bin": {

7 "cawcut": "dist/index.js"

8 },

9 "main": "dist/index.js",

10 "files": [

11 "dist",

12 "config",

13 "assets"

14 ],

15 "scripts": {

16 "build:ts": "node scripts/build.mjs",

17 "build": "npm run build:ts",

18 "build:internal": "node scripts/build.mjs --profile internal",

19 "prepare": "npm run build",

20 "postinstall": "node dist/postinstall.js",

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L20

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg · L20

21 "reinstall:global": "npm uninstall -g @ubnt/cawcut && npm install -g . --foreground-scripts --allow-scripts=@ubnt/cawcut",

22 "test": "vitest run",

23 "verify:package": "node scripts/verify-npm-package.mjs",

24 "release:check": "npm run test && npm run build && npm run verify:package",

25 "prepublishOnly": "node scripts/assert-release-ci.mjs && npm run release:check"

26 },

27 "dependencies": {

28 "commander": "^13.1.0"

29 },

30 "devDependencies": {

31 "@types/node": "^22.15.0",

32 "typescript": "^5.8.3",

33 "vitest": "^3.1.4"

34 },

35 "engines": {

36 "node": ">=20"

37 },

38 "license": "MIT"

39}

dist/lib/vn-media.jsView file

Lines 1-21javascript

1import { createHash } from "node:crypto";

2import { execFile } from "node:child_process";

3import { copyFile, mkdir, open, readFile, stat } from "node:fs/promises";

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/lib/vn-media.jsView on unpkg · L1

dist/lib/skills-agents.jsView file

Lines 1-21javascript

1import { existsSync } from "node:fs";

2import { homedir } from "node:os";

Critical

Ai Agent Control Hijack

Source creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.

dist/lib/skills-agents.jsView on unpkg · L1

3import { join } from "node:path";

dist/lib/vn-pack.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 2

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/lib/vn-pack.js#virtual:normalized:round1View on unpkg

dist/lib/is-global-install.jsView file

•matchType = normalized_sha256 matchedPackage = @ubnt/cawcut@0.1.5 matchedPath = dist/lib/is-global-install.js matchedIdentity = npm:QHVibnQvY2F3Y3V0:0.1.5 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

dist/lib/is-global-install.jsView on unpkg

Source & flagged code

6 flagged · loading source

package.jsonView file

•scripts.postinstall = node dist/postinstall.js

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = node dist/postinstall.js

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg

dist/lib/vn-media.jsView file

1import { createHash } from "node:crypto"; L2: import { execFile } from "node:child_process"; L3: import { copyFile, mkdir, open, readFile, stat } from "node:fs/promises"; ... L419: async function probeWithPlatformTool(filePath) { L420: if (process.platform === "darwin") { L421: try { L422: const { stdout } = await execFileAsync("/usr/bin/afinfo", [filePath], { timeout: 10_000 }); L423: const match = stdout.match(/estimated duration:\s*([\d.]+)\s*sec/i);

Low

Weak Crypto

Findings

1 Critical3 High4 Medium6 Low

CriticalAi Agent Control Hijackdist/lib/skills-agents.js

HighInstall Time Lifecycle Scriptspackage.json

HighSemantic Analysis Limiteddist/lib/vn-pack.js#virtual:normalized:round1

HighKnown Malware Source Similaritydist/lib/is-global-install.js

MediumAmbiguous Install Lifecycle Scriptpackage.json

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

LowWeak Cryptodist/lib/vn-media.js

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to @ubnt/cawcut@0.6.0.

See all versions of @ubnt/cawcut for other recorded verdicts.

Review the evidence and your use of @ubnt/cawcut@0.6.0 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: 2026-10-12 21:10:36.179Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.