Native filesystem for agents, editors, and cloud sandboxes

Cloud storage and APIs, mounted like local files.

Parlo creates a real ~/parlo folder on macOS and Linux. Cloud drives stream byte ranges on demand and cache hot data locally; API connections appear as markdown files your shell, editor, scripts, and AI agents already know how to use.

parlo — zsh
mounted

$ ls ~/parlo

drive/ github/ linear/ projects/ stripe/

$ ls ~/parlo/projects/video

q3-launch.mov timeline.prproj

$ head -3 ~/parlo/linear/ENG-123.md

--- id: ENG-123 state: In Progress

Parlo Drive
Humans
$ curl -fsSL https://parlo.run/install | sh
Agents
$ curl -fsSL https://parlo.run/agent | sh -s -- --yes

Works with your favorite tools & platforms

Claude CodeClaude Code
Codex CLICodex CLI
CursorCursor
AiderAider
OpenCodeOpenCode
WindsurfWindsurf
ContinueContinue
GooseGoose
Zed AIZed AI
DevinDevin
CodyCody
Replit AgentReplit Agent
e2be2b
ModalModal
DaytonaDaytona
VercelVercel
CloudflareCloudflare
RunloopRunloop
Claude CodeClaude Code
Codex CLICodex CLI
CursorCursor
AiderAider
OpenCodeOpenCode
WindsurfWindsurf
ContinueContinue
GooseGoose
Zed AIZed AI
DevinDevin
CodyCody
Replit AgentReplit Agent
e2be2b
ModalModal
DaytonaDaytona
VercelVercel
CloudflareCloudflare
RunloopRunloop
Built for agents that already know files

Let agents explore the workspace instead of loading every tool.

Parlo mounts APIs and cloud storage as native folders and files. Agents discover with ls, inspect with cat, and search with grep — so they act on only the exact resource needed.

Without Parlo100 MCP tools loaded

6,638 prompt tokens · 6 tool calls

With Parlo~/parlo mounted

1,407 prompt tokens · 1 file read

78.8% less prompt context · same result

Connect anything. Then use it however.

Watch a new service get mounted live. Once it's connected, you'll see how to use it.

Connect any APIparlo connect …

One filesystem. Three interfaces. Same Stripe API calls under the hood.

APIs become folders. Storage becomes a drive.

Use pre-validated services or paste any API docs. Parlo turns resources into files and keeps large cloud media thin until something actually reads bytes.

GitHubGitHub
LinearLinear
StripeStripe
SlackSlack
NotionNotion
VercelVercel
OpenAIOpenAI
AnthropicAnthropic
SupabaseSupabase
CloudflareCloudflare
SentrySentry
FigmaFigma
IntercomIntercom
HubSpotHubSpot
ShopifyShopify
AsanaAsana
JiraJira
DatadogDatadog
TwilioTwilio
SendGridSendGrid
SalesforceSalesforce
AirtableAirtable
Any API

Paste any domain — parlo connect example.com — and the wizard generates the adapter on the fly.

Edit a 38 GB project without downloading 38 GB.

Parlo Drive mounts cloud storage as a real folder. Premiere, Logic, your shell — they read what they need, when they need it. The other 99% never touches your SSD.

No bulk sync

Mount first. Fetch later. Parlo only downloads the files and byte ranges your app actually asks for.

Local hot cache

Hot files and scrubbed video ranges stay on local disk, so repeat opens and timeline revisits are effectively instant.

Agent-ready workspace

Agents can list folders, read specific files, patch changes, and avoid pulling an entire drive into context.

Share with a link

parlo share ~/parlo/projects/q3 — get a /s/:id URL anyone can open, no account needed.

Why Parlo is different

Parlo is not another dashboard or SDK. It is a native filesystem surface for cloud storage, APIs, and agents.

Native mounts

Parlo supports macOS and Linux as first-class native drive targets. Finder, cp, editors, and apps see normal paths instead of a web dashboard or custom client.

parlo connect linear

APIs as markdown

Issues, customers, docs, jobs, and internal resources become .md files with frontmatter. Humans and agents can inspect and patch them with the same tools.

parlo connect <anything>

On-demand cloud drive

Large files stream by range, cache locally, and sync across devices. You mount the workspace first; bytes move only when something reads or writes them.

cat ~/parlo/x/y.md

Sandbox ready

The same CLI works on macOS, Linux, and supported cloud sandbox/container environments such as Cloudflare Sandbox/Containers. Good for local humans and remote agents.

parlo mount

One mount. Three audiences.

Creative apps, developers, and AI agents all operate on the same mounted paths.

Humans

Open cloud footage from a native drive. Edit a Linear issue in vim. Browse project folders in Finder. The tools you already use become the interface.

See the CLI
AI agents

Claude Code, Cursor, Aider, and sandboxed agents can list, read, patch, and write through normal paths. They already speak filesystem.

Agent workflow
Scripts

Cron jobs and shell scripts use cp, cat, grep, mv, and rm. No SDK glue just to move data between storage, APIs, and automation.

Automate it

How it works

Install, connect, mount, use. Each step is one command.

  1. 01

    Install

    npm install -g parlofs, or use the hosted installer. Works on macOS and Linux with native drive support.

  2. 02

    Connect

    Pick a service or paste any domain. The wizard generates the filesystem adapter in ~10s and tests it against your token.

  3. 03

    Mount

    parlo mount — drives and API connections appear under ~/parlo/ as normal folders.

  4. 04

    Use

    Use Finder, cp, open, ls, cat, vim, grep, or an AI agent. Parlo translates file operations to storage and API calls.

FAQ