Why Use Tenki?

What is Tenki?

Tenki is a developer platform for shipping code with confidence — AI code review for pull requests, drop-in GitHub Actions runners, and disposable Linux microVMs for AI agents. It is operated by Luxor, a profitable, SOC 1 and SOC 2 Type II certified company founded in 2017 and based in Seattle, Washington.

• Three first-class products. Code Reviewer for PR decisions, Runners for CI, Sandbox for AI agents.
• Drop-in for GitHub. Runners need one YAML change; Code Reviewer installs as a GitHub App.
• Per-job isolation. Every workload runs in its own ephemeral Firecracker microVM, destroyed when it ends.
• Usage-based pricing. Per-minute billing, $10/month free credits on the Starter plan, no credit card required.
• Compliance posture. SOC 1 and SOC 2 Type II via Luxor.

See Security & Isolation for the full compliance posture.

Why Tenki Code Reviewer?

Tenki Code Reviewer is an AI reviewer that posts inline comments, suggested edits, and a summary directly on the pull request — like a teammate, not a dashboard you have to visit.

Built for merge decisions, not nitpicks

The reviewer surfaces merge-blocking risk and high-impact issues, and stays silent when there's nothing worth flagging. You spend review time on what matters.

Follows your team's conventions

Wire in custom context — coding standards, architectural rules, naming conventions, things you don't want flagged — and the reviewer adapts to your codebase rather than a generic style guide. Severity and verbosity are tunable per repository.

No agent to run, no workflow to add

Code Reviewer is a GitHub App. Install it on your org or selected repositories, and reviews start on the next pull request — typical setup takes under two minutes with no infrastructure to operate.

Reviews at PR speed

Reviews run as soon as a pull request is opened or updated, so contributors get first-pass feedback before a human reviewer is paged. Re-trigger anytime by tagging @tenki-reviewer in a PR comment.

See the Code Reviewer quickstart to install it.

Why Tenki Runners?

Tenki runners are a drop-in replacement for GitHub-hosted runners: change one value in your workflow YAML (the runs-on field) and your existing Actions jobs run unmodified on faster, cheaper hardware.

Secure

Every job runs in its own ephemeral Firecracker microVM that boots in around 15 seconds on average, regardless of runner size or concurrency. The VM is destroyed at the end of the workflow — jobs cannot inspect or interfere with one another, and job-time secrets never leave the VM boundary.

We operate our own dedicated compute pool, so your CI jobs never compete with noisy neighbors or unpredictable on-demand workloads like they would on AWS or Hetzner. Running on ephemeral VMs we control eliminates third-party supply-chain risk.

Scalable

Autoscaling is automatic. Configure your runs-on value once — our Migration Wizard does it in two clicks — and Tenki adjusts runner capacity based on each job's actual needs. No pre-allocated pool, no idle capacity to pay for, and no guesswork about whether you want 2, 4, or 8 vCPUs.

Cheaper

Tenki offers a compelling price-to-performance advantage over GitHub-hosted runners:

• Up to 10x more cost-effective for common CI workloads.
• Fair, simple, per-minute billing with no startup surcharge and no minimum commit.
• $10 in monthly free credits on the Starter plan, renewed every month, no credit card required.

Standard runner pricing

• Tenki standard runner (2 vCPU / 4 GB): $0.003 per minute.
• GitHub-hosted Linux 2-core: $0.008 per minute (public GitHub pricing).
• Savings on the standard profile: ~62%.

See the pricing page for the full rate card including higher vCPU, memory, and macOS profiles.

Performance

Tenki runners are powered by modern bare-metal hardware that outpaces the older VM instances GitHub Actions uses, with shorter build times and a smoother CI experience. Lower per-minute cost also means upgrading to more powerful runners (4+ cores) often still saves money while accelerating builds.

Benchmarks vs GitHub-hosted runners

Workload	GitHub-hosted	Tenki	Delta
Rust cargo build	5s	3s	40% faster
Docker build	27s	19s	30% faster
Node.js npm install	10s	8s	20% faster
Go build	11s	0.1s	99% faster
Android assembleDebug	1m 38s	1m 2s	37% faster
n8n monorepo (full CI)	55m 58s	29m 15s	48% faster

Source: Tenki internal benchmarks, run against identical GitHub Actions workflows on ubuntu-latest vs tenki-standard-medium-4c-8g with warm cache. Full methodology and raw logs available on request at [email protected].

See the Runners quickstart to migrate.

Why Tenki Sandbox?

Tenki Sandbox provides disposable Linux microVMs built for AI agents and untrusted code. Every agent task runs in its own isolated workspace — agents write, run, and ship code safely without touching your host machine.

Zero blast radius

Every session runs in an isolated Firecracker microVM with zero cross-contamination between workloads. Code, dependencies, and secrets never leak across boundaries; when the session ends, the VM is destroyed.

Sub-100ms cold starts

Sandboxes provision a live environment in under 100 milliseconds, with no cold-start image pulls between sessions. That latency budget is what makes sandboxes usable inside an agent loop — you can spin up a fresh environment between tool calls without making the user wait.

Hundreds of sandboxes, one API call

Spin up hundreds of concurrent sandboxes for batch workloads, eval harnesses, or multi-agent orchestration, each fully isolated with dedicated resources. There is no shared scheduler bottleneck and no noisy-neighbor effect across sessions.

Drive from anywhere

Pick the surface that fits your workflow: the CLI for terminal flows, the TypeScript or Go SDK for application code, or the Sandbox IDE desktop app for agent-driven sessions. All three speak to the same isolated microVMs.

See the Sandbox quickstart to spin up your first session.