registry / AI-agent capability abuse / @rubytech/create-sitedesk-code / 0.1.629

@rubytech/create-sitedesk-code@0.1.629

Install SiteDesk — automated back office for independent building contractors

AI Security Review

scanned 1mo ago · by lpm-firewall-ai

There is no npm install-time hook or self-dependency chain; Claude writes are brand-scoped and scanner exfil labels did not match a hidden drop. Residual risk is the explicit installer seeding bypassPermissions, persisting services, and deleting ~/.claude on uninstall.

Static reason

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

Trigger

The user runs the create-sitedesk-code binary, including via npx, or the same binary with --uninstall.

Impact

Unattended Claude sessions skip permission prompts; uninstall can destroy existing Claude CLI login state; the host gains services, sudoers, crontab, and a global git URL rewrite.

Mechanism

Explicit-command first-party Claude and host provisioning, with uninstall-time wipe of default Claude state

AI rationale

Evidence

package.jsondist/index.jsdist/permissions-seed.jsdist/uninstall.js~/.sitedesk-code/.claude/settings.json~/.sitedesk-code/.claude/.claude.json~/.claude~/.ollama/etc/sudoers.d/maxy-samba~/sitedesk-code/

Network endpoints4

registry.npmjs.org/github.com/api.telegram.orgapi.cloudflare.com

Decision evidence

public snapshot

AI called this Suspicious at 86.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

Running the create-sitedesk-code CLI immediately provisions the host: system packages, Claude Code, Neo4j, user services, cron, Samba sudoers, and a hardware watchdog, with no install confirmation prompt.
The installer seeds Claude Code defaultMode bypassPermissions into brand-scoped settings so unattended sessions skip permission checks.
Uninstall deletes the user's default ~/.claude directory (Claude CLI OAuth state) and ~/.ollama unless another Maxy-family brand is still present.
The installer rewrites the user's global git URL mapping for GitHub.
Linux installs merge a brand cron block into the user crontab and register systemd or launchd persistence.

Evidence against

package.json has no preinstall, install, or postinstall scripts; the only lifecycle script is author-side prepublishOnly.
Claude plugin and settings writes use CLAUDE_CONFIG_DIR under ~/.sitedesk-code/.claude rather than the default ~/.claude/settings.json.
There is no runtime dependency on this package name, so no nested self-install lifecycle chain.
Telegram, Cloudflare, Google, npm, and GitHub usage matches the SiteDesk product, not a hidden credential drop.

Source & flagged code

31 flagged

payload/platform/scripts/smoke-boot-services.shView file

349patternName = generic_password severity = medium line = 349

Medium

Secret Pattern

Package contains a possible secret pattern.

payload/platform/scripts/smoke-boot-services.shView on unpkg · L349

dist/index.jsView file

1#!/usr/bin/env node L2: import { execFileSync, spawn, spawnSync } from "node:child_process"; L3: import { existsSync, mkdirSync, writeFileSync, cpSync, readFileSync, rmSync, readdirSync, appendFileSync, openSync, closeSync, chmodSync, statSync, realpathSync, accessSync, consta... ... L9: import { buildBrandCronBlock, mergeBrandCronBlock } from "./cron-registration.js"; L10: import { validateTierFlag, validateEntitlementBase64, applyTierToAccountConfig, entitlementPath } from "./tier-flag.js"; L11: import { parseOsRelease, isUbuntuLike as isUbuntuLikePure, parseAptCacheCandidate, decideAptResolution, } from "./apt-resolve.js"; ... L52: } L53: const INSTALL_DIR = resolve(process.env.HOME ?? "/root", BRAND.installDir); L54: const PERSIST_DIR = resolve(process.env.HOME ?? "/root", BRAND.co

Critical

Download Execute

Source downloads or fetches remote code and executes it.

dist/index.jsView on unpkg · L1

1Trigger-reachable chain: manifest.bin -> dist/index.js L1: #!/usr/bin/env node L2: import { execFileSync, spawn, spawnSync } from "node:child_process"; L3: import { existsSync, mkdirSync, writeFileSync, cpSync, readFileSync, rmSync, readdirSync, appendFileSync, openSync, closeSync, chmodSync, statSync, realpathSync, accessSync, consta... ... L9: import { buildBrandCronBlock, mergeBrandCronBlock } from "./cron-registration.js"; L10: import { validateTierFlag, validateEntitlementBase64, applyTierToAccountConfig, entitlementPath } from "./tier-flag.js"; L11: import { parseOsRelease, isUbuntuLike as isUbuntuLikePure, parseAptCacheCandidate, decideAptResolution, } from "./apt-resolve.js"; ... L52: } L53: const INSTALL_DIR = resolve(process.env.HOME ?? "/root", BRAND.installDir); L54: const PERS

Critical

Trigger Reachable Dangerous Capability

A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.

dist/index.jsView on unpkg · L1

High

Child Process

Package source references child process execution.

dist/index.jsView on unpkg · L1

537// Fetch + dearmor the signing key, write to /etc/apt/trusted.gpg.d/. Pipe L538: // composition runs through bash -c so the curl|gpg pipeline is one L539: // privileged command rather than two separate sudo escalations.

High

Shell

Package source references shell execution.

dist/index.jsView on unpkg · L537

High

Sandbox Evasion Gated Capability

Source gates dangerous network, credential, or execution behavior behind CI, host, platform, time, or geo fingerprint checks.

dist/index.jsView on unpkg · L1

1Trigger-reachable persistence chain: manifest.bin -> dist/index.js L1: #!/usr/bin/env node L2: import { execFileSync, spawn, spawnSync } from "node:child_process"; L3: import { existsSync, mkdirSync, writeFileSync, cpSync, readFileSync, rmSync, readdirSync, appendFileSync, openSync, closeSync, chmodSync, statSync, realpathSync, accessSync, consta... ... L9: import { buildBrandCronBlock, mergeBrandCronBlock } from "./cron-registration.js"; L10: import { validateTierFlag, validateEntitlementBase64, applyTierToAccountConfig, entitlementPath } from "./tier-flag.js"; L11: import { parseOsRelease, isUbuntuLike as isUbuntuLikePure, parseAptCacheCandidate, decideAptResolution, } from "./apt-resolve.js"; ... L52: } L53: const INSTALL_DIR = resolve(process.env.HOME ?? "/root", BRAND.installDir); L54

High

Trigger Reachable Persistence

A manifest entrypoint or package-local install chain reaches persistence behavior.

dist/index.jsView on unpkg · L1

285function canSudo() { L286: const result = spawnSync("sudo", ["-n", "true"], { stdio: "pipe", timeout: 5_000 }); L287: return result.status === 0; ... L289: /** True when the npm global prefix's node_modules dir is writable by the current L290: * user (nvm/Homebrew installs) — meaning `npm install -g` needs no sudo. */ L291: function npm[redacted]() {

High

Runtime Package Install

Package source invokes a package manager install command at runtime.

dist/index.jsView on unpkg · L285

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = dist/index.js matchedIdentity = npm:[redacted]:0.1.632 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/index.jsView on unpkg

Medium

Install Persistence

Source writes installer persistence such as shell profile or service configuration.

dist/index.jsView on unpkg · L1

payload/platform/plugins/ledger/lib/mcp-spawn-tee/index.jsView file

8* server: Claude Code runs `node <this> <real-entry>`, and the shim runs the L9: * real entry IN ITS OWN PROCESS via dynamic import() — one node runtime, not L10: * two. Before importing, it replaces process.stderr.write with a tee so every

Medium

Dynamic Require

Package source references dynamic require/import behavior.

payload/platform/plugins/ledger/lib/mcp-spawn-tee/index.jsView on unpkg · L8

payload/platform/plugins/admin/skills/flow-page/build.mjsView file

99const LOG_FILE = L100: process.env.LOG_DIR && existsSync(process.env.LOG_DIR) L101: ? join(process.env.LOG_DIR, "flow-page.log") ... L178: let model; L179: try { model = JSON.parse(readFileSync(contentPath, "utf8")); } L180: catch (e) { die(`[flow-page] op=bad-content slug=${slug} file=${contentPath} error=${e.message}`); } ... L209: FONT_DISPLAY: fonts.display, L210: FONT_BODY: fonts.body, L211: ACCENT: palette.accent,

Low

Weak Crypto

Package source references weak cryptographic algorithms.

payload/platform/plugins/admin/skills/flow-page/build.mjsView on unpkg · L99

payload/server/server.jsView file

358try { L359: out.push(JSON.parse(line)); L360: } catch { ... L469: function telegramStoreRoot2(platformRoot5) { L470: const brandDir = process.env.MAXY_DIR_OVERRIDE ?? (0, node_path_1.resolve)((0, node_os_1.homedir)(), configDirName2(platformRoot5)); L471: return (0, node_path_1.join)(brandDir, "data", "telegram-messages"); ... L566: var node_path_1 = __require("path"); L567: var TELEGRAM_API2 = "https://api.telegram.org"; L568: function refusalReason(body, status) { ... L742: }); L743: process.stderr.write(`[trash:marked] accountId=${accountId} elementId=${elementId} labels=${baseLabels.join(",")} by=${by} reason=${reason ?? "null"} L744: `);

Critical

Credential Exfiltration

Source appears to send environment or credential material to an external endpoint.

payload/server/server.jsView on unpkg · L358

5370// app/lib/vnc.ts L5371: import { spawnSync, execFileSync } from "child_process"; L5372: import { createConnection } from "net"; L5373: import { mkdirSync as mkdirSync2, readFileSync as readFileSync4, rmSync as rmSync2, writeFileSync as writeFileSync3 } from "fs"; L5374: import { resolve as resolve5 } from "path"; L5375: var PLATFORM_ROOT2 = process.env.MAXY_PLATFORM_ROOT ?? resolve5(process.cwd(), ".."); L5376: var VNC_SCRIPT = resolve5(PLATFORM_ROOT2, "scripts/vnc.sh");

High

Same File Env Network Execution

A single source file combines environment access, network access, and code or shell execution; review context before blocking.

payload/server/server.jsView on unpkg · L5370

payload/platform/plugins/admin/mcp/dist/index.jsView file

6Cross-file remote execution chain: [redacted].js spawns dist/index.js; helper contains network access plus dynamic code execution. L6: import { resolve, join } from "node:path"; L7: import { execFile, execFileSync } from "node:child_process"; L8: import { appendFileSync, cpSync, existsSync, mkdirSync, readdirSync, readFileSync, renameSync, statSync, writeFileSync } from "node:fs"; ... L19: import { createHash, randomInt, randomUUID } from "node:crypto"; L20: import { createConnection } from "node:net"; L21: import { homedir, hostname as osHostname } from "node:os"; ... L61: }); L62: const PLATFORM_ROOT = process.env.PLATFORM_ROOT ?? resolve(import.meta.dirname, "../../../.."); L63: // Plugin-system boot tolerance — see Task 202. ... L67: const ACCOUNT_ID = process.env.ACCOUNT_ID ?? null; L

High

Cross File Remote Execution Context

Source spawns a local helper that also contains network and dynamic execution context; review data flow before blocking.

payload/platform/plugins/admin/mcp/dist/index.jsView on unpkg · L6

payload/platform/plugins/memory/bin/conversation-archive-ingest.shView file

•path = [redacted]-archive-ingest.sh kind = build_helper sizeBytes = 4369 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

payload/platform/plugins/memory/bin/conversation-archive-ingest.shView on unpkg

payload/server/public/assets/graph-labels-BTFUSc63.js.gzView file

•path = [redacted]-labels-BTFUSc63.js.gz kind = high_entropy_blob sizeBytes = 9968 magicHex = [redacted]

High

Ships High Entropy Blob

Package ships high-entropy non-source blobs.

payload/server/public/assets/graph-labels-BTFUSc63.js.gzView on unpkg

•path = [redacted]-labels-BTFUSc63.js.gz kind = compressed_blob sizeBytes = 9968 magicHex = [redacted]

Medium

Ships Compressed Blob

Package ships compressed or archive-like blobs.

payload/server/public/assets/graph-labels-BTFUSc63.js.gzView on unpkg

payload/platform/lib/account-enumeration/dist/index.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 44

High

Semantic Analysis Limited

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

payload/platform/lib/account-enumeration/dist/index.js#virtual:normalized:round1View on unpkg

dist/uninstall.jsView file

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = dist/uninstall.js matchedIdentity = npm:[redacted]:0.1.632 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/uninstall.jsView on unpkg

payload/platform/lib/mcp-spawn-tee/dist/index.jsView file

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = [redacted]-spawn-tee/index.js matchedIdentity = npm:[redacted]:0.1.632 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.

payload/platform/lib/mcp-spawn-tee/dist/index.jsView on unpkg

payload/platform/plugins/admin/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/admin/lib/mcp-spawn-tee/index.jsView on unpkg

payload/platform/plugins/admin/skills/_lib/headless.mjsView file

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = [redacted]_lib/headless.mjs matchedIdentity = npm:[redacted]:0.1.632 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.

payload/platform/plugins/admin/skills/_lib/headless.mjsView on unpkg

payload/platform/plugins/admin/skills/video-production/render.mjsView file

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = [redacted]-production/render.mjs matchedIdentity = npm:[redacted]:0.1.632 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.

payload/platform/plugins/admin/skills/video-production/render.mjsView on unpkg

payload/platform/plugins/aeo/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/aeo/lib/mcp-spawn-tee/index.jsView on unpkg

payload/platform/plugins/browser/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/browser/lib/mcp-spawn-tee/index.jsView on unpkg

payload/platform/plugins/cloudflare/bin/portal-index-push.mjsView file

•matchType = normalized_sha256 matchedPackage = @rubytech/create-maxy-code@0.1.632 matchedPath = [redacted]-index-push.mjs matchedIdentity = npm:[redacted]:0.1.632 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.

payload/platform/plugins/cloudflare/bin/portal-index-push.mjsView on unpkg

payload/platform/plugins/connector/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/connector/lib/mcp-spawn-tee/index.jsView on unpkg

payload/platform/plugins/contacts/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/contacts/lib/mcp-spawn-tee/index.jsView on unpkg

payload/platform/plugins/dispatch/lib/mcp-spawn-tee/index.jsView file

High

Known Malware Source Similarity

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

payload/platform/plugins/dispatch/lib/mcp-spawn-tee/index.jsView on unpkg

Source & flagged code

31 flagged · loading source

payload/platform/scripts/smoke-boot-services.shView file

349patternName = generic_password severity = medium line = 349

Medium

Secret Pattern

Package contains a possible secret pattern.

payload/platform/scripts/smoke-boot-services.shView on unpkg · L349

dist/index.jsView file

Critical

Download Execute

Source downloads or fetches remote code and executes it.

dist/index.js

Findings

3 Critical22 High8 Medium9 Low

CriticalCredential Exfiltrationpayload/server/server.js

CriticalDownload Executedist/index.js

CriticalTrigger Reachable Dangerous Capabilitydist/index.js

HighChild Processdist/index.js

HighShelldist/index.js

HighSame File Env Network Executionpayload/server/server.js

HighCommand Output Exfiltrationpayload/server/server.js

HighSandbox Evasion Gated Capabilitydist/index.js

HighCross File Remote Execution Contextpayload/platform/plugins/admin/mcp/dist/index.js

HighTrigger Reachable Persistencedist/index.js

HighRuntime Package Installdist/index.js

Affected versions and remediation

This report applies to @rubytech/create-sitedesk-code@0.1.629.

See all versions of @rubytech/create-sitedesk-code for other recorded verdicts.

Review the evidence and your use of @rubytech/create-sitedesk-code@0.1.629 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: 2026-09-16 19:05:31.616Z (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

Related packages

More from @rubytech

Behavioral surface

Related packages

More from @rubytech

Source & flagged code