Parlo

Start

What is Parlo?

Parlo is a filesystem interface for cloud files and APIs, built for humans, scripts, and agents.

Parlo mounts cloud drives and APIs as files.

After you connect a service, it appears under ~/parlo like a normal folder. Humans, scripts, editors, and agents can use commands they already know: ls, cat, grep, cp, vim, rm, and normal file reads/writes.

bash6 lines
1~/parlo/2  main/        # cloud-backed files3  github/      # GitHub issues, PRs, repos, docs4  linear/      # Linear issues and projects5  stripe/      # Stripe customers, invoices, subscriptions6  search/      # an API connector you named search

The exact folder names depend on what you connect and what you name each mount.

A real use case

Say you want to investigate a billing bug. Without Parlo, an agent may need separate GitHub tools, Linear tools, Stripe tools, Drive tools, and a lot of tool descriptions in context.

With Parlo, the same task is just filesystem work:

bash34 lines
1parlo mount2cd ~/parlo3 4ls5# main  github  linear  stripe6 7cat github/README.md8cat linear/README.md9 10# Find the related product and support records without reading everything.11grep -R "billing\|invoice\|subscription" github/issues linear/issues | head -2012cat github/issues/42.md13cat linear/issues/ENG-123.md14 15# Check the customer record exposed by the Stripe connector.16cat stripe/customers/cus_8x92.md17 18# Leave a report in a cloud-backed drive.19mkdir -p main/reports20cat > main/reports/billing-bug.md <<'EOF'21# Billing bug notes22 23GitHub issue:24- github/issues/42.md25 26Linear ticket:27- linear/issues/ENG-123.md28 29Affected customer:30- stripe/customers/cus_8x92.md31 32Likely cause:33- subscription status cache is stale after Stripe webhook success34EOF

The important part is not the specific providers. It is that GitHub issues, Linear tickets, Stripe customers, and cloud-drive files all become paths that humans, scripts, and agents can use.

Each API mount also explains itself with a README.md:

bash14 lines
1cat github/README.md2 3# Example output:4# # GitHub5#6# Useful paths:7# - issues/       issue files8# - pulls/        pull request files9# - repos/        repository metadata10#11# Examples:12#   ls issues/13#   cat issues/42.md14#   grep -R "billing" issues/

That is the core idea: APIs and drives become files, so the same shell/file behavior works across systems.

What Parlo is for

NeedParlo gives you
Work with APIsmarkdown files backed by real API calls
Store and share filescloud drives that mount like folders
Open large mediastreaming reads and local chunk cache
Give agents accessa small file interface instead of huge tool catalogs
Search big reposa separate remote grep feature for codebases

The model

Parlo has one namespace:

bash4 lines
1~/parlo/2  main/                 # cloud drive3  github/               # API connection4  linear/               # API connection

Different backends, one interface.

Each mount has its own README.md. Read that first. It tells you what folders exist, which files are read-only, and which paths can be edited.

bash2 lines
1cat ~/parlo/linear/README.md2cat ~/parlo/github/README.md

Remote repo grep is a separate feature for code search. Use parlo rg when you want to search a large GitHub repo without cloning it first.

bash2 lines
1parlo rg "TODO" owner/repo2parlo rg malloc -t cpp v8/v8

Coming soon: compute workspaces

The same model can extend to compute: user-named folders for servers, GPU jobs, logs, and artifacts. Provider names are metadata, not necessarily mount names. A user might create ~/parlo/training/ backed by RunPod, Modal, DigitalOcean, Hetzner, Kubernetes, or another provider.

bash6 lines
1~/parlo/training/2  provider.md3  machines/4  jobs/5  logs/6  artifacts/

See Compute mounts.

Two ways to use it

Use the native mount when FUSE/native mounting is available:

bash4 lines
1parlo mount2cd ~/parlo3ls4cat github/README.md

Use direct commands for CI, containers, and agent sandboxes:

bash3 lines
1parlo ls github/issues/2parlo cat linear/issues/ENG-123.md3parlo write main/reports/billing-report.md --from /tmp/billing-report.md

Where to go next

GoalRead
Install ParloQuickstart
Understand the path modelFilesystem model
Connect an APIAPI connections
Use cloud files and streamingCloud drives
Search codebases with remote grepCodebase search
Use Parlo from an agentAgent access
Explore the compute roadmapCompute mounts
Next →

Quickstart