Jern Cloud

jern.ai

Jern Cloud is a governed cloud coding agent that runs each session on an isolated machine under a repository-pinned policy, with hard token caps and encrypted evidence, then hands back a pull request for human review — secure, credential-free, and open-source under Apache-2.0.

Overview

Jern Cloud is a governed cloud coding agent launched in September 2026 that reframes autonomous coding around trust and reviewability rather than maximum autonomy. You install its GitHub App on selected repositories, open a session, and describe the work; Jern then runs each message as one attempt on a fresh isolated machine under a policy pinned for the whole session, with a hard token cap and encrypted evidence, and finally opens a pull request for human review. Nothing runs on your laptop, and critically no credential ever reaches the agent process — checkout, evidence upload, model inference, and publication each use a separate short-lived credential held by the runner, whose only network route is the inference gateway. The repository’s baseline decides which paths can change and which commands run without asking; the runner verifies that digest before starting and it cannot change for the life of the session. Tokens are counted by the gateway from the provider’s own response, and wall-clock, process count, output size, and attempts are all capped. Every attempt’s exact trace is encrypted with AES-256-GCM before reaching the database and can be opened from any assistant turn, with retention you control. Work ends on an isolated branch as a pull request your branch protection and reviewers govern — Jern never merges. It is open-source under Apache-2.0 and supports bring-your-own Anthropic, OpenAI, Gemini, or Fireworks keys.

Key Features

  • Policy pinned by digest — allowed paths and commands verified before the agent starts and frozen for the session
  • No credential in the agent — separate short-lived credentials for checkout, evidence, inference, and publish; only the inference gateway is reachable
  • Hard, gateway-metered caps — tokens counted from the provider’s own response, plus wall-clock, process, output, and attempt limits
  • Encrypted evidence — every attempt trace AES-256-GCM encrypted, retrievable, with retention you set
  • Immutable environment — image and setup command pinned by revision; mid-session environment changes are impossible
  • Review-only exit — changes land as a pull request on an isolated branch; Jern never merges or deploys

Pricing

PlanPriceFor
Free trial$030 days of full access per workspace
Workspace$19/moPer workspace after trial; up to 10 repos and team members, BYOK
RuntimeApache-2.0Self-host the open-source agent runtime

Comparison

Compared to GitHub Copilot, Jern removes the agent from your machine and replaces inline autonomy with repository-pinned policy and encrypted evidence, which reviewers can audit after the fact. Against Claude Code, Jern’s sessions are isolated and credential-free by construction, at the cost of being cloud-run rather than local. If you want a local, open-source coding agent you fully control, OpenCode is the closer fit, but it does not provide Jern’s governed, PR-only workflow out of the box.

Compare alternatives

Side-by-side with the 3 closest alternatives.

ToolCategoryPricingVisit
Jern Cloud (this) code, agentsFree $0 · From $19/mo Site ↗
GitHub CopilotcodeFrom $10/mo Site ↗
Claude Codecode, agentsFrom $20/mo Site ↗
OpenCodecode, agentsFree $0 Site ↗
Jern Cloud Current

Jern Cloud is a governed cloud coding agent that runs each session on an isolated machine under a repository-pinned policy, with hard token caps and encrypted evidence, then hands back a pull request for human review — secure, credential-free, and open-source under Apache-2.0.

codeagents
Free $0 · From $19/mo

An AI coding assistant by GitHub and OpenAI that delivers real-time code completion inside your editor. Fast, inline completions in VS Code and JetBrains

code
From $10/mo

Anthropic's agentic coding CLI that writes, runs, and ships code directly from your terminal, IDE, Slack, or the web using Claude models.

codeagents
From $20/mo

OpenCode is an open-source AI coding agent that runs in your terminal, desktop, or IDE, connects to 75+ LLM providers including local models, and offers a Claude Code-grade experience with no vendor lock-in or subscription.

codeagents
Free $0
Editor’s Review
4.3/5
Pros
  • +Governed by a repository-pinned policy file: allowed paths and commands are verified by digest and cannot change mid-session
  • +No credential in the agent — checkout, evidence, inference, and publication each use separate short-lived credentials; the only network route is the inference gateway
  • +Encrypted, reviewable evidence for every attempt (AES-256-GCM) plus hard token, wall-clock, and attempt caps
Cons
  • −Runs in the cloud, not on your laptop, so code leaves your environment during the session
  • −Bring-your-own-key only (Anthropic, OpenAI, Gemini, Fireworks) — there is no managed model bundled in
  • −Young project (launched September 2026) with a small community and limited third-party validation so far

Jern Cloud is a refreshing take on coding agents: instead of chasing autonomy, it makes the session auditable and bounded, so a reviewer can trust the pull request without supervising every step. The policy-pinning and encrypted-evidence model is genuinely useful for teams that need proof, not vibes — just accept that it is cloud-run and BYOK.

See all reviews →