Skills/npm/parallelsandbox-mcp

parallelsandbox-mcp@0.7.4

CAUTION

Publisher claims: "Remote Linux boxes for coding agents: Docker, a browser, screenshots, logs, human takeover." Observed in sandbox: calls mcp.parallelsandbox.com only, touches nothing outside the project, writes no memory.

Registry npm (MCP server) · listed in the official MCP RegistryPublisher parallelsandbox · publishing for 60 daysInstalls 0Last detonated 3d ago
What would fix thisWatchDispute verdict

Coverage

Server started and answered tools/list.
Tested
MCP initialize and tools/list · every tool description, for hidden instructions · every file read, file written and host contacted by what ran
Not tested
what the tools do when called with real credentials

Before you give it your token

It will ask for
API_KEYGEMINI_API_KEYPARALLELSANDBOX_API_KEYPARALLELSANDBOX_AUTH_DIRConfiguration, not credentials: CLAUDE_CODE_HOST_SESSION_ID, CLAUDE_CODE_SESSION_ID, CLAUDE_CONFIG_DIR, CLAUDE_PID, CLAUDECODE, CODEX_HOME, GEMINI_MODEL, GOOGLE_GENAI_USE_VERTEXAI, PARALLELSANDBOX_API_URL, PARALLELSANDBOX_MCP_URL, PARALLELSANDBOX_NO_BROWSER, PSBX_ADAPTER_ANSWER_MS, PSBX_ADAPTER_NO_CONNECT, PSBX_ADAPTER_NOT_SENT_DELAY_MS, PSBX_ADAPTER_SILENCE_MS, PSBX_ADAPTER_STALL_MS, PSBX_ADAPTER_TOOLS_CHECK_MS, PSBX_ADAPTER_UNAVAILABLE_BUDGET_MS, PSBX_ADAPTER_UNAVAILABLE_FIRST_MS, PSBX_CLAUDE_ALLOWED_TOOLS, PSBX_CLAUDE_MODEL, PSBX_CODEX_ALLOWED_MCP_TOOLS, PSBX_CODEX_APPROVAL_POLICY, PSBX_CODEX_CLI, PSBX_CODEX_FEEDBACK_DIR, PSBX_CODEX_HOST_SOCKET, PSBX_CODEX_MODEL, PSBX_CODEX_REASONING_EFFORT, PSBX_CODEX_SANDBOX_MODE, PSBX_FEEDBACK_DIR, PSBX_FEEDBACK_HOST, PSBX_GEMINI_APPROVAL_MODE, PSBX_HOOK_HOLD_SEC, PSBX_SESSION_SOCKET, PSBX_TOOL_TIMEOUT_SEC
Where it talks to
parallelsandbox.com (its own)
What the honeypot saw
no credential files read · reached only mcp.parallelsandbox.com (its own)
What we could not test
what the tools do when called with real credentials

Connect it with a scoped credential, under the guard. It asks for API_KEY, GEMINI_API_KEY, PARALLELSANDBOX_API_KEY, PARALLELSANDBOX_AUTH_DIR and talks to parallelsandbox.com. Give the credential to the guard, not the server: the value never enters the server's environment and only travels to the host you bind.

$vet-skills run --secrets API_KEY=parallelsandbox.com,GEMINI_API_KEY=parallelsandbox.com,PARALLELSANDBOX_API_KEY=parallelsandbox.com,PARALLELSANDBOX_AUTH_DIR=parallelsandbox.com -- npx -y parallelsandbox-mcp

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios) on 2026-11-13 and scored it 69/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Access to sensitive file paths; Destructive file system command detected; Access to sensitive file paths. Summary: calls mcp.parallelsandbox.com only, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 26 s · 2 scenarios
t+CategoryObserved action
0.0sprocessno install scripts in bundle
6.0sprocessdependency install exit 0
26sfilesystemread ~/.parallelsandbox/oauth-5c345124870d6c3479a70664f05c6206.json.lock · ~/.parallelsandbox/oauth-5c345124870d6c3479a70664f05c6206.json · ~/.cache/parallelsandbox/feedback/tickets outside project
26sfilesystemwrote ~/.parallelsandbox/oauth-5c345124870d6c3479a70664f05c6206.json.lock/94 · ~/.parallelsandbox/oauth-5c345124870d6c3479a70664f05c6206.json.94.a786645d7168.tmp
26snetworkCONNECT tcp://mcp.parallelsandbox.com:443
26snetworkCONNECT tcp://mcp.parallelsandbox.com:443
26sprocessexec /app/apps/web/node_modules/.bin/xdg-open
26sprocessexec /root/.cache/node/corepack/v1/pnpm/10.28.0/dist/node-gyp-bin/xdg-open
26sprocessexec /app/node_modules/.bin/xdg-open
26sprocessexec /pnpm/xdg-open
26sprocessexec /usr/local/sbin/xdg-open
26sprocessexec /usr/local/bin/xdg-open
26sprocessexec /usr/sbin/xdg-open
26sprocessexec /usr/bin/xdg-open
26sprocessexec /sbin/xdg-open
26sprocessexec /bin/xdg-open
26sclaimed1 tool: parallelsandbox_connect
26stools8ddf03d5a8b8f95b
26sprocessscript backend runs setup only
26sprocessno install scripts to re-run

Static findings

MEDIUM
File permission modification commandindex.mjs:1506 · File permission modification command — agent changes file ownership/permissions…), so ./${first} fails with Permission denied in the box: run bash ${first}, or chmod +x it there. To fix it in the repo: git update-index --chmod=+x ${first}.`; } /…Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
HIGH
Access to sensitive file pathsindex.mjs:58 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesconst RELOADS_DEV_SERVER = /^(package\.json|vite\.config\..+|tsconfig.*|\.env.*)$/;Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
npx invocation of non-well-known packagefeedback-hook.mjs:132 · npx invocation of non-well-known package — may execute untrusted code// (npx and a shell may sit in between).Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
LOW
Fake instruction boundary tagREADME.md:187 · Fake instruction boundary tag · informational: not an instruction file…riginal conversation is not running and asks the person to do that. `claude -p "<prompt>"` runs the hook in the foreground, so there the call that opens a review waits …Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
HIGH
Destructive file system command detectedindex.mjs:1359 · Destructive file system command detectednotes.push(`/work/${target.rel} is already a folder in the box, so the file went inside it. If an older adapter made that folder by mistake, remove it (rm -r /work/${target.rel}) and sync again.`);Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
HIGH
Access to sensitive file pathsindex.mjs:929 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files// configs 是被忽略的檔裡看起來是建置設定的（package.json、tsconfig*.json、*.config.*、.env*），整個被忽略的資料夾裡的不算。Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
HIGH
Access to sensitive file pathsindex.mjs:800 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files// dropIgnored 用本機 repo 的忽略規則篩掉 rels（相對 src 的路徑 Buffer）裡被忽略的：箱子裡的 .env、建置產物本來就不從本機來，Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
HIGH
Access to sensitive file pathsnative-gemini.mjs:69 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesenv: { ...env, GEMINI_CLI_HOME: nativeHome },Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
HIGH
Access to sensitive file pathsindex.mjs:64 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesconst BUILD_CONFIG = /^(package\.json|tsconfig.*\.json|jsconfig\.json|.+\.config\.[^.]+|\.env.*)$/;Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
HIGH
Access to sensitive file pathsnative-claude.mjs:112 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesenv: { ...env, CLAUDE_CONFIG_DIR: configDir },Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
File type combined with upload actionfeedback-agent.mjs:38 · File type combined with upload action — agent may exfiltrate local data filesif (existsSync(join(stateDir, "config.json"))) throw new Error("Session already exists; use send with its exact --session path");Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
HIGH
Access to sensitive file pathsnative-codex.mjs:68 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesconst childEnv = { ...env, CODEX_HOME: stateDir };Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
Process kill commandindex.mjs:430 · Process kill command — agent can terminate processes// kill -9 來不及打 leave，control 三分鐘沒收到心跳也會當它走了。Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagefeedback-install.mjs:11 · npx invocation of non-well-known package — may execute untrusted code// The command that runs this same package: npx for npx installs (its cache path is not stable),Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
HIGH
Access to sensitive file pathsindex.mjs:1316 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files// devServerNote：送過去的檔裡有 package.json、vite.config.*、tsconfig*、.env* 時，提醒開著的 dev server 會整頁重載或重啟。Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
npx invocation of non-well-known packagefeedback-install.mjs:14 · npx invocation of non-well-known package — may execute untrusted codereturn /[\\/]_npx[\\/]/.test(index) ? "npx -y parallelsandbox-mcp" : `"${node}" "${index}"`;Fix: Pin versions and sources; vendor anything fetched at install time.
AST06

Cost to the agent

Instructions loaded
≈75 tok
1 file · 24 in bundle
Tool list per prompt
≈69 tok
1 tool · 69 per tool
Handshake
465 ms
launch to tools/list, stdio
Undocumented tools
0
description under 20 chars · avg 172

Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.

Known vulnerabilities

No advisories in OSV for parallelsandbox-mcp@0.7.4 as of 3d ago. This is the package itself, not its dependencies.

What would make this Verified

11 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%
  1. Unclaimed publisher≈ +8 on the totalClaim parallelsandbox: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
  2. Touches files outside the project≈ +7 on the totalKeep reads and writes inside the working directory, or declare the exact paths in `filesystem:` and explain why in the README.
  3. Access to sensitive file paths≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
  4. Destructive file system command detected≈ +4 on the totalBuild commands from fixed arguments; do not interpolate untrusted strings into a shell.
  5. File permission modification command≈ +2 on the totalRequest the narrowest permission that works; replace wildcards with the exact paths or hosts.
  6. npx invocation of non-well-known package≈ +2 on the totalPin versions and sources; vendor anything fetched at install time.
Publish a new version to npm, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim parallelsandboxPropose a patch to the maintainer