Loading npm security reports…

devez-vibe@1.7.9: AI-agent control-surface risk | LPM Firewall

registry / devez-vibe / 1.7.9

devez-vibe@1.7.9

Stable terminal UI for Codex and Claude Agent SDK

AI Security Review

scanned 3h ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installing the package automatically replaces skills in the user's Codex and Claude control directories. Its bridge also attempts to use Claude permission bypass.

Static reason

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

Trigger

npm installation runs postinstall.

Impact

Package-authored instructions can persist in two user agent environments without an explicit setup command.

Mechanism

Automatic cross-agent skill deployment and permission-bypass preference.

Policy narrative

On installation, npm launches install-skills.mjs. It copies the bundled luna-loop skill into both the user's Codex and Claude skill locations and prunes files there, creating an unconsented persistent cross-agent control-surface change. The bundled bridge additionally attempts to select bypassPermissions for Claude sessions. No network theft was found, but the automatic broad agent mutation is concrete.

Rationale

This is an unconsented postinstall mutation of foreign AI-agent control surfaces, compounded by code that prefers permission bypass. The behavior meets the install-control-surface block policy even without observed exfiltration.

Evidence

package.jsoninstall-skills.mjsbridge/claude-agent-sdk-bridge.mjsskills/luna-loop/$CODEX_HOME/skills/luna-loop$CLAUDE_CONFIG_DIR/skills/luna-loop

Decision evidence

Source & flagged code

7 flagged · loading source

package.jsonView file

•scripts.postinstall = node install-skills.mjs

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

40"scripts": { "postinstall": "node install-skills.mjs" }

High

Suspicious Lifecycle Evidence

The npm postinstall hook automatically runs the skill installer.

package.jsonView on unpkg · L40

install-skills.mjsView file

1import { L2: copyFileSync, L3: existsSync, L4: mkdirSync, L5: readdirSync, ... L14: const userHome = homedir(); L15: const codexHome = process.env.CODEX_HOME?.trim() || join(userHome, ".codex"); L16: const claudeHome = process.env.CLAUDE_CONFIG_DIR?.trim() || join(userHome, ".claude"); L17: ... L23: function copyTree(source, destination) { L24: mkdirSync(destination, { recursive: true }); L25: for (const entry of readdirSync(source, { withFileTypes: true })) {

Critical

Ai Agent Control Hijack

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

install-skills.mjsView on unpkg · L1

12const packageRoot = dirname(fileURLToPath(import.meta.url)); const sourceRoot = join(packageRoot, "skills", "luna-loop"); const userHome = homedir(); const codexHome = process.env.CODEX_HOME?.trim() || join(userHome, ".codex"); const claudeHome = process.env.CLAUDE_CONFIG_DIR?.trim() || join(userHome, ".claude"); const targets = [ { name: "Codex", path: join(codexHome, "skills", "luna-loop") }, { name: "Claude", path: join(claudeHome, "skills", "luna-loop") }, ];

High

Ai Review Evidence

The installer copies package-controlled files into both Codex and Claude user skill directories.

install-skills.mjsView on unpkg · L12

bin/dvz.exeView file

•path = bin/dvz.exe kind = native_binary sizeBytes = 24389632 magicHex = [redacted]

Medium

Ships Native Binary

Package ships native binary artifacts.

bin/dvz.exeView on unpkg

bridge/claude-agent-sdk-bridge.mjsView file

•matchType = normalized_sha256 matchedPackage = devez-vibe@1.7.1 matchedPath = bridge/claude-agent-sdk-bridge.mjs matchedIdentity = npm:ZGV2ZXotdmliZQ:1.7.1 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.

bridge/claude-agent-sdk-bridge.mjsView on unpkg

•matchType = malicious_source_fingerprint_signature signature = 7ad776c0d0eeac91 signatureType = suspicious_hashes sourceLabel = final_verdict:malicious matchedPackage = devez-vibe@1.7.1 matchedPath = bridge/claude-agent-sdk-bridge.mjs matchedIdentity = npm:ZGV2ZXotdmliZQ:1.7.1 similarity = 1.000 shingleOverlap = 2 summary = package final verdict is malicious

High

Known Malware Source Fingerprint Signature

Source fingerprint signature matches a known malicious package signature; route for source-aware review.

bridge/claude-agent-sdk-bridge.mjsView on unpkg

Source & flagged code

7 flagged

package.jsonView file

Lines 20-47json

20 "bin": {

21 "dvz": "bin/dvz.exe"

22 },

23 "files": [

24 "bin/dvz.exe",

25 "bridge/claude-agent-sdk-bridge.mjs",

26 "install-skills.mjs",

27 "skills/luna-loop/",

28 "README.md",

29 "LICENSE"

30 ],

31 "os": [

32 "win32"

33 ],

34 "cpu": [

35 "x64"

36 ],

37 "engines": {

38 "node": ">=18"

39 },

40 "scripts": {

41 "postinstall": "node install-skills.mjs"

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L41

42 },

High

Suspicious Lifecycle Evidence

The npm postinstall hook automatically runs the skill installer.

package.jsonView on unpkg · L40

43 "dependencies": {

44 "@anthropic-ai/claude-agent-sdk": "0.3.251"

45 }

46}

47

install-skills.mjsView file

Lines 1-32javascript

1import {

2 copyFileSync,

3 existsSync,

4 mkdirSync,

5 readdirSync,

6 rmSync,

7} from "node:fs";

8import { homedir } from "node:os";

bin/dvz.exeView file

•path = bin/dvz.exe kind = native_binary sizeBytes = 24389632 magicHex = [redacted]

Medium

Ships Native Binary

Package ships native binary artifacts.

bin/dvz.exeView on unpkg

bridge/claude-agent-sdk-bridge.mjsView file

•matchType = normalized_sha256 matchedPackage = devez-vibe@1.7.1 matchedPath = bridge/claude-agent-sdk-bridge.mjs matchedIdentity = npm:ZGV2ZXotdmliZQ:1.7.1 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.

bridge/claude-agent-sdk-bridge.mjsView on unpkg

•matchType = malicious_source_fingerprint_signature signature = 7ad776c0d0eeac91 signatureType = suspicious_hashes sourceLabel = final_verdict:malicious matchedPackage = devez-vibe@1.7.1 matchedPath = bridge/claude-agent-sdk-bridge.mjs matchedIdentity = npm:ZGV2ZXotdmliZQ:1.7.1 similarity = 1.000 shingleOverlap = 2 summary = package final verdict is malicious

High

Known Malware Source Fingerprint Signature

Source fingerprint signature matches a known malicious package signature; route for source-aware review.

bridge/claude-agent-sdk-bridge.mjsView on unpkg