Tanvrit Compute
Reserve early access
Quickstart

Hello, Compute — in 5 minutes.

This walks through installing the CLI and the desktop agent, authenticating, and running a hello-world job. Steps 1-5 work today against the local hobby tier. The cloud-tier endpoints are stubbed until public alpha — running this against --target cloud will return ServiceUnavailable until then.

Local hobby tier only · cloud-tier returns 503 until alpha invites
01

Install the CLI

One-line installer detects your OS / arch and drops tanvrit-compute on $PATH. macOS, Linux, WSL2, x86_64 + arm64.

input
curl -L https://compute.tanvrit.com/install.sh | sh
output
Installing tanvrit-compute v0.1.0 (darwin/arm64)...
Downloaded https://artifacts.tanvrit.com/compute/0.1.0/darwin-arm64/tanvrit-compute
Verified sha256 ... ok
Installed to /usr/local/bin/tanvrit-compute
Run 'tanvrit-compute --help' to get started.

Brew formula and Windows MSI / Linux DEB packages ship in phase 2 — see CLI reference.

02

Install the desktop agent

The agent is the local executor. It picks up jobs from your workspace, runs them under your user account, and reports status back to the CLI.

See /agents for the full installer (DMG / MSI / DEB). Or run from source via the CLI:

input
tanvrit-compute agent run --workspace local
output
agent: starting workspace=local mode=embedded
agent: registered with control plane (local)
agent: ready — listening for jobs
03

Authenticate

Browser-based device-code OAuth flow. Stores the token in your OS keychain (macOS Keychain / Windows DPAPI / libsecret on Linux). The agent reads it from the same place.

input
tanvrit-compute auth login
output
Open this URL in your browser:
  https://compute.tanvrit.com/auth/device?code=BX-9K2-PQM
Waiting for confirmation... ok
Logged in as you@example.com (workspace: local)
Token cached in: macOS Keychain (com.tanvrit.compute.cli)
04

Run hello-world

Drop this YAML into hello-world.yaml and submit it. The local agent picks it up, runs it under your user, and streams logs back.

hello-world.yaml
# hello-world.yaml
apiVersion: compute.tanvrit.com/v0
kind: Job
metadata:
  name: hello-world
spec:
  type: COMMAND
  command: echo "hello from tanvrit-compute"
  timeoutSeconds: 30
input
tanvrit-compute jobs run hello-world.yaml
output
Submitting hello-world.yaml ...
● Submitted   job_01jt9hello · queued → assigned → running
● Streaming logs from agent local-01
  hello from tanvrit-compute
✓ SUCCESS    0.42s · exit 0
05

View the result

Inspect the job by id — status, exit code, total runtime, and a link to the full log buffer.

input
tanvrit-compute jobs status job_01jt9hello
output
id:        job_01jt9hello
state:     SUCCEEDED
exit:      0
runtime:   0.42s
agent:     local-01
workspace: local
logs:      tanvrit-compute jobs logs job_01jt9hello

What next?