registry / AI-agent capability abuse / gm-plugkit / 2.0.2522

gm-plugkit@2.0.2522

Bootstrap and daemon-spawn tool for gm plugkit binary. Downloads the correct platform wasm, verifies SHA256, and launches agentplug-runner (the native wasm host) as the spool watcher daemon.

AI Security Review

scanned 3mo ago · by lpm-firewall-ai

Source confirms an explicit-user-command agent configuration and remote-code supply-chain capability. No lifecycle hook, credential theft, or autonomous exfiltration was found, so this should warn rather than block.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

User runs gm-plugkit (for example via bun x/npx) or calls ensureReady.

Impact

Can influence future AI-agent behavior across projects and execute remotely supplied binaries.

Mechanism

Remote agent-skill installation, project instruction wiring, and downloaded runner execution.

AI rationale

Evidence

package.jsoncli.jsbootstrap.jsbootstrap-shared.jsSKILL.mdCLAUDE.mdAGENTS.md.gm/next-step.md.gm/instructions/~/.agents/skills/<skill>/SKILL.md~/.claude/skills/<skill>/SKILL.md~/.gm-tools/agentplug-runner

Network endpoints3

api.github.com/repos/AnEntrypoint/gm/contents/skills?ref=mainraw.githubusercontent.com/AnEntrypoint/gm/main/skills/<skill>/SKILL.mdgithub.com/AnEntrypoint/agentplug-bin/releases/download/<tag>/<asset>

Decision evidence

public snapshot

AI called this Suspicious at 98.0% confidence as Dangerous Capability with low false-positive risk.

Evidence for warning

bootstrap.js:624 refreshes skills into ~/.agents/skills and ~/.claude/skills.
bootstrap.js:636 discovers remote skills; missing local skills are fetched from GitHub main.
bootstrap.js:803 invokes project wiring on every explicit bootstrap.
bootstrap-shared.js:266 prepends CLAUDE.md and appends AGENTS.md.
SKILL.md directs agents to replace questions/summaries with its command protocol.
cli.js:145 executes the downloaded agentplug-runner.

Evidence against

package.json has no preinstall/install/postinstall hooks.
Execution and writes require an explicit CLI/bootstrap invocation.
No credential harvesting or outbound data-exfiltration logic found.
Release artifacts are SHA-256 checked, though hashes are fetched remotely.

Behavioral surface

Source

ChildProcessCryptoEnvironmentVarsFilesystemNetwork

Supply chain

HighEntropyStringsUrlStrings

Source & flagged code

3 flagged

bootstrap-shared.jsView file

Lines 1-25javascript

1'use strict';

3const fs = require('fs');

4const path = require('path');

5const os = require('os');

6const { spawnSync } = require('child_process');

7const { pidAlive, sha256OfFileSync } = require('./gm-process');

High

Child Process

Package source references child process execution.

bootstrap-shared.jsView on unpkg · L5

10const LOCK_STALE_MS = 30 * 60 * 1000;

11const ATTEMPT_TIMEOUT_MS = 10 * 60 * 1000;

13function obsEvent(subsystem, event, fields) {

14 if (process.env.GM_LOG_DISABLE) return;

15 try {

16 const root = process.env.GM_LOG_DIR || path.join(os.homedir(), '.claude', 'gm-log');

17 const day = new Date().toISOString().slice(0, 10);

18 const dir = path.join(root, day);

19 fs.mkdirSync(dir, { recursive: true });

20 const line = JSON.stringify({

21 ts: new Date().toISOString(),

22 sub: subsystem,

23 event,

24 pid: process.pid,

25 sess: process.env.CLAUDE_SESSION_ID || process.env.GM_SESSION_ID || '',

bootstrap.jsView file

Lines 1-27javascript

1#!/usr/bin/env node

2'use strict';

4const fs = require('fs');

5const path = require('path');

Source & flagged code

3 flagged · loading source

bootstrap-shared.jsView file

5const os = require('os'); L6: const { spawnSync } = require('child_process'); L7: const { pidAlive, sha256OfFileSync } = require('./gm-process');

High

Child Process

Package source references child process execution.

bootstrap-shared.jsView on unpkg · L5

bootstrap.jsView file

7const crypto = require('crypto'); L8: const { spawn, spawnSync } = require('child_process'); L9: const { sha256OfFile, sha256OfFileSync } = require('./gm-process'); ... L37: L38: const wrapperDir = __dirname; L39: ... L44: try { L45: const ownPkg = JSON.parse(fs.readFileSync(path.join(__dirname, 'package.json'), 'utf-8')); L46: if (ownPkg && ownPkg.version) { ... L167: const dir = gmToolsDir(); L168: const names = process.platform === 'win32' L169: ? ['agentplug-runner.exe']

Critical

Download Execute

Source downloads or fetches remote code and executes it.

bootstrap.jsView on unpkg · L7

7Trigger-reachable chain: manifest.main -> index.js -> bootstrap.js L7: const crypto = require('crypto'); L8: const { spawn, spawnSync } = require('child_process'); L9: const { sha256OfFile, sha256OfFileSync } = require('./gm-process'); ... L37: L38: const wrapperDir = __dirname; L39: ... L44: try { L45: const ownPkg = JSON.parse(fs.readFileSync(path.join(__dirname, 'package.json'), 'utf-8')); L46: if (ownPkg && ownPkg.version) { ... L167: const dir = gmToolsDir(); L168: const names = process.platform === 'win32' L169: ? ['agentplug-runner.exe']

Findings

2 Critical1 High3 Medium3 Low

CriticalDownload Executebootstrap.js

CriticalTrigger Reachable Dangerous Capabilitybootstrap.js

HighChild Processbootstrap-shared.js

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to gm-plugkit@2.0.2522.

See all versions of gm-plugkit for other recorded verdicts.

Review the evidence and your use of gm-plugkit@2.0.2522 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: 2026-08-07 19:32:24.287Z (UTC).

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

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code