Files
paperclip/packages/plugins/sandbox-providers/kubernetes/src/job-orchestrator.ts
T
Jannes Stubbemann 05ab45225a feat(plugin-kubernetes): self-hostable Kubernetes sandbox provider (stage 1/3: plugin package) (#5790)
## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - Sandbox providers are the seam that lets agent runs execute in
isolated environments; today the only first-party remote provider is
Daytona, a hosted third-party service
> - Self-hosters running Paperclip on their own infrastructure (often
Kubernetes already) have no first-party way to run agent sandboxes on a
cluster they control
> - That gap matters for teams with data-residency, sovereignty, or cost
constraints who cannot or will not send workloads to a hosted sandbox
service
> - This pull request adds a Kubernetes sandbox-provider plugin as a
standalone, workspace-excluded package: it implements every
SandboxProvider hook the Daytona provider does, on infrastructure the
operator owns
> - The benefit is that any Paperclip deployment with a Kubernetes
cluster gets multi-tenant, network-isolated, quota-bounded agent
sandboxes with zero new external dependencies

## Linked Issues or Issue Description

No existing issue. Following the feature template:

- **Problem:** Paperclip's remote sandbox execution requires a hosted
third-party provider. Self-hosters cannot run agent sandboxes on their
own Kubernetes clusters with a first-party provider.
- **Proposed solution:** A `@paperclipai/plugin-kubernetes`
sandbox-provider plugin with two backends: long-lived sandboxes via the
[kubernetes-sigs/agent-sandbox](https://github.com/kubernetes-sigs/agent-sandbox)
CRD (multi-command exec, adapter-install pattern) and one-shot
`batch/v1` Jobs (stable APIs only, no extra controllers).
- **Alternatives considered:** Driving kubectl from a generic shell
provider (no lifecycle/lease semantics), or requiring a hosted provider
(exactly the constraint this removes).

## What Changed

This is **stage 1 of 3** of a staged contribution (direction agreed with
maintainers): the plugin package alone. Stage 2 (server integration:
lease params, provider registration) and stage 3 (agent runtime images +
CI) are companion PRs that will be cross-linked from a comment here.

- New package `packages/plugins/sandbox-providers/kubernetes`
(workspace-excluded, like the path already carved out in
`pnpm-workspace.yaml`): src, unit + kind integration tests, operator
prerequisite manifests, README, smoke-test guide
- Implements the full SandboxProvider hook surface the Daytona provider
implements: `validateConfig`, `probe`, `acquireLease`, `resumeLease`,
`releaseLease`, `destroyLease`, `realizeWorkspace`, `execute`
- Two backends: `sandbox-cr` (default; long-lived pod via the
agent-sandbox `Sandbox` CR, supports multi-command exec) and `job`
(one-shot `batch/v1` Job; nothing beyond k8s 1.27+ required)
- Per-run adapter resolution: one environment serves mixed harnesses;
the per-run `adapterType` hint is read through a local optional type
extension, so the plugin typechecks and builds against the current
plugin SDK and simply falls back to the environment's configured default
adapter until stage 2 lands
- Exec-env wrapping: the Kubernetes exec API carries no environment, so
commands are wrapped to receive the run's env
- Fast-upload interception for workspace realization, scoped per lease
- Per-tenant isolation: derived namespace per company, RBAC,
ResourceQuota, restricted-PSS pod security (runAsNonRoot, drop ALL,
seccomp RuntimeDefault, no SA token automount)
- Network egress policy in two flavors: native `NetworkPolicy` and
`CiliumNetworkPolicy` (FQDN allowlists)
- Image allowlist with glob matching, registry override, and per-run
image override validation
- Per-run Kubernetes Secrets carrying agent credentials, ownerRef'd to
the Job or Sandbox CR for cascade GC

## Verification

- Standalone build, exactly as the README documents:
  ```bash
  cd packages/plugins/sandbox-providers/kubernetes
  pnpm install --ignore-workspace
  pnpm test        # 147 unit tests, 17 files, all green
  pnpm typecheck   # clean against the in-repo plugin SDK on master
pnpm build # dist/ emitted, manifest + worker entrypoints present
  ```
- A kind-cluster end-to-end integration test is included
(`RUN_K8S_INTEGRATION_TESTS=1 pnpm test
test/integration/end-to-end-run.test.ts`)
- Beyond CI: this provider has been verified in a production
multi-tenant deployment against five harnesses (opencode, pi, codex,
gemini, claude code) with real billed runs

## Risks

- **Zero behavior change for any existing deployment.** The package is
workspace-excluded; nothing in the server imports or loads it until
stage 2's integration lands. No existing code paths are touched.
- The default `sandbox-cr` backend depends on an alpha CRD
(`agents.x-k8s.io/v1alpha1`); the README flags this and the `job`
backend uses only stable APIs as a fallback.
- Risk surface is confined to deployments that explicitly install and
configure the plugin.
- The default runtime images (`ghcr.io/paperclipai/agent-runtime-*`) are
published by the stage 3 companion PR (#7934); until that lands,
deployments must point `runtimeImage` at their own images.

## Model Used

Claude Opus 4.8 (1M context), extended thinking, with tool use (Claude
Code).

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots (no UI changes)
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green (pending this push)
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-06-10 21:07:00 -07:00

126 lines
4.6 KiB
TypeScript

import type { KubeClients } from "./kube-client.js";
import type { SandboxOrchestrator, SandboxStatus } from "./sandbox-orchestrator.js";
export class JobTimeoutError extends Error {
constructor(namespace: string, name: string, timeoutMs: number) {
super(`Job ${namespace}/${name} did not complete within ${timeoutMs}ms`);
this.name = "JobTimeoutError";
}
}
export async function createJob(
clients: KubeClients,
namespace: string,
manifest: Record<string, unknown>,
): Promise<{ uid: string }> {
const result = await clients.batch.createNamespacedJob({ namespace, body: manifest as never });
const uid = (result as { metadata?: { uid?: string } }).metadata?.uid;
if (!uid) throw new Error("Job created without a UID");
return { uid };
}
export type JobStatus = SandboxStatus;
export async function getJobStatus(
clients: KubeClients,
namespace: string,
name: string,
): Promise<JobStatus> {
const result = await clients.batch.readNamespacedJobStatus({ namespace, name });
const body = (result as Record<string, unknown>) ?? {};
const status = (body.status as Record<string, unknown>) ?? {};
const active = (status.active as number) ?? 0;
const succeeded = (status.succeeded as number) ?? 0;
const failed = (status.failed as number) ?? 0;
const conditions = (status.conditions as { type: string; status: string; reason?: string; message?: string }[]) ?? [];
const completed = conditions.find((c) => c.type === "Complete" && c.status === "True");
const failedCond = conditions.find((c) => c.type === "Failed" && c.status === "True");
if (failedCond || failed > 0) {
return { phase: "Failed", complete: false, active, succeeded, failed, reason: failedCond?.reason, message: failedCond?.message };
}
if (completed || succeeded > 0) {
return { phase: "Succeeded", complete: true, active, succeeded, failed };
}
if (active > 0) {
return { phase: "Running", complete: false, active, succeeded, failed };
}
return { phase: "Pending", complete: false, active, succeeded, failed };
}
export async function findPodForJob(
clients: KubeClients,
namespace: string,
jobName: string,
): Promise<string | null> {
const result = await clients.core.listNamespacedPod({
namespace,
labelSelector: `job-name=${jobName}`,
});
const items = ((result as { items?: { metadata?: { name?: string }; status?: { phase?: string } }[] }).items) ?? [];
const running = items.find((p) => p.status?.phase === "Running");
return (running ?? items[0])?.metadata?.name ?? null;
}
export async function streamPodLogs(
clients: KubeClients,
namespace: string,
podName: string,
onChunk: (stream: "stdout" | "stderr", text: string) => Promise<void>,
): Promise<void> {
// V1 limitation: readNamespacedPodLog returns combined stdout (the kubectl-style
// log view). stderr is not separately exposed via this API path — agent
// containers that need stderr/stdout separation should use a sidecar log
// scraper or wrap their CLI to emit structured output on stdout. We always
// emit chunks as "stdout"; the "stderr" callback slot in SandboxOrchestrator
// is unused by the Job-backed implementation.
const result = await clients.core.readNamespacedPodLog({ namespace, name: podName });
const text = (result as string) ?? "";
if (text.length > 0) await onChunk("stdout", text);
}
export async function deleteJob(
clients: KubeClients,
namespace: string,
name: string,
): Promise<void> {
await clients.batch.deleteNamespacedJob({
namespace,
name,
propagationPolicy: "Foreground",
});
}
export async function waitForJobCompletion(
clients: KubeClients,
namespace: string,
name: string,
opts: { timeoutMs: number; pollMs?: number } = { timeoutMs: 120_000, pollMs: 2000 },
): Promise<JobStatus> {
const deadline = Date.now() + opts.timeoutMs;
const pollMs = opts.pollMs ?? 2000;
while (Date.now() < deadline) {
const status = await getJobStatus(clients, namespace, name);
if (status.phase === "Succeeded" || status.phase === "Failed") return status;
await sleep(pollMs);
}
throw new JobTimeoutError(namespace, name, opts.timeoutMs);
}
function sleep(ms: number): Promise<void> {
return new Promise((resolve) => setTimeout(resolve, ms));
}
/**
* Job-backed conformance to SandboxOrchestrator. Plugin.ts imports THIS value
* (the swap point) — to use a different backend, swap this import for another
* module exposing a SandboxOrchestrator-shaped default export.
*/
export const jobOrchestrator: SandboxOrchestrator = {
claim: createJob,
getStatus: getJobStatus,
findPod: findPodForJob,
streamLogs: streamPodLogs,
release: deleteJob,
waitForCompletion: waitForJobCompletion,
};