ParallelSandbox MCP server
online · 100% uptimecom.parallelsandbox/parallelsandbox·v1.0.18Remote Linux boxes for coding agents: Docker, a browser, screenshots, logs, human takeover.
not scanned yet
364 downloads/wk
Reviews
Write oneNobody has reviewed ParallelSandbox yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
ParallelSandbox tools
Tool list not cached yet. `describe` through the gateway fetches it live.
How ParallelSandbox compares
Against every graded Scraping and browsing MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 72/100
- higher than 38% of 866 Scraping and browsing servers
- Uptime
- 100%
- of our recent checks; the typical remote Scraping and browsing server answers 100%
- Response time
- 735 ms
- faster than 12% of 384 Scraping and browsing servers
- Downloads a week
- 364
- more than 81% of 309 Scraping and browsing servers with a package
Release history: 13 versions in the registry since 2026-09-24, latest 2026-10-27, typically one every 2 days.
Install ParallelSandbox in Claude Code, Cursor or VS Code
claude mcp add --transport http parallelsandbox https://mcp.parallelsandbox.com/mcp
What the publisher says
From the ParallelSandbox repository's README, as published. We do not edit it. Read it on GitHub
parallelsandbox-mcp
stdio adapter for the ParallelSandbox MCP server. It exposes the same tools as https://mcp.parallelsandbox.com/mcp (sandbox, logs), forwarding every call with your API key, and implements sandboxsync and sandboxpull locally: an agent can upload its working directory into a box without committing, and write files from the box back to its machine. Get an API key (psbx_…) in the app: https://app.parallelsandbox.com → Account → API keys.
Use
PARALLELSANDBOX_API_KEY=psbx_... npx parallelsandbox-mcpClaude Code:
claude mcp add parallelsandbox -e PARALLELSANDBOX_API_KEY=psbx_... -- npx -y parallelsandbox-mcpAny other MCP client: run the command above as a stdio server.
If you never need sandboxsync or sandboxpull, you can skip this package and connect directly (over plain HTTP those two tools return an error, because they read and write files on your machine):
claude mcp add --transport http parallelsandbox https://mcp.parallelsandbox.com/mcp --header "Authorization: Bearer psbx_..."Environment
sandbox_sync
{"id": "", "localPath": "/abs/path/to/repo", "dest": "repo"} syncs localPath into /work/ on the box. Pass an absolute path: a relative one resolves against this adapter's working directory (where your MCP client started it), which is not necessarily the agent's. In a git working tree it takes the tracked files plus the untracked files .gitignore does not exclude, asks the box which of them differ from what is already in dest, and uploads only those, so syncing after an edit sends just that edit. A file changed inside the box counts as different and gets the local version back. Outside git it uploads the whole directory, excluding node_modules, .git, dist, build, coverage, .venv and similar. Files deleted locally stay in the box.
sandbox_pull
{"id": "", "path": "", "localPath": "/abs/path/on/your/machine"} downloads a file or a directory from the box. A directory arrives as a tar.gz and is extracted into localPath (pass "extract": false to keep the archive); a file is written to localPath. Relative paths resolve the same way as for sandbox_sync.
Which conversation is using a box
Each time the adapter starts it picks a random id and sends it as X-Psbx-Agent on every call. After the first tool call it reports to /v1/agents//heartbeat once a minute, and when its conversation closes (stdin ends, or SIGTERM, SIGINT or SIGHUP) it reports /v1/agents//leave. The app uses this to tell a box the AI is still working in from one whose conversation was closed without sandboxreview or sandboxstop; a conversation killed outright (no leave) counts as gone three minutes after its last heartbeat. Nothing else is sent: no prompt, no transcript.
Timeouts
sandboxtakeover blocks until a person hands the box back, up to 30 minutes; the adapter waits 31 minutes for it and 65 minutes for other calls (a foreground sandboxexec runs at most 60 minutes).
Shortened. The full README is on GitHub.
Nothing above is checked by us. What we check is on the safety report.
ParallelSandbox: common questions
- Is ParallelSandbox MCP server safe?
- Mostly: it is graded B (72/100). Read the ParallelSandbox safety report
- How do I install ParallelSandbox?
- It runs remotely at mcp.parallelsandbox.com. Add it to Claude Code, Cursor, VS Code or Claude Desktop with the snippets above, or call it through the mcp.market gateway without installing anything.
- Does ParallelSandbox need an API key?
- Yes. The registry entry asks for
PARALLELSANDBOX_API_KEY,Authorization. - Is ParallelSandbox maintained?
- The latest release is v1.0.18.
- Is ParallelSandbox up?
- 100% of our last 28 checks got an answer. We check remote servers about four times a day.
- What can I use instead of ParallelSandbox?
- Servers from other publishers that do the same job: Hwatu MCP server, GrowBlocks Dev Kit MCP server and Flyto Core MCP server.
- How does ParallelSandbox compare with other Scraping and browsing MCP servers?
- safety score 72/100 (higher than 38% of 866 Scraping and browsing servers); uptime 100% (of our recent checks; the typical remote Scraping and browsing server answers 100%); response time 735 ms (faster than 12% of 384 Scraping and browsing servers). Full comparison.
Alternatives to ParallelSandbox
Same job from other publishers: the closest match first, then the best rated.
- HwatuVerification browser for coding agents: headless WebKit, DOM eval, screenshots, pixel-diff.not reviewedGrowingA
- GrowBlocks Dev KitBuild, play-test with screenshots, and ship three.js browser games on GrowBlocks.5.0(1)NewA
- Flyto CoreExecution engine for AI agents. 412 modules: browser, file, Docker, data, crypto.not reviewedEstablishedB
- BrowserlessOfficial Browserless MCP server: browser automation, scraping, screenshots and PDFs for AI agents.not reviewedEstablishedB
- Vessel BrowserAI-native browser runtime for autonomous agents with human-in-the-loop supervisionnot reviewedEstablishedA