~/reading

36

Issue 36/October 9, 2026/8 tools · 2 headlines

MicroVM control planes and cloud-native harnesses outpace chat wrappers

Today's picks center on programmable isolation layers and replaceable agent loops that power users can clone, wire, and run under their own policies. Sandbox-as-tool patterns and durable session control planes dominate over single-vendor IDEs.

01 — Power tools

The stack

  1. 01repoGitHub
    Gondolin↗

    earendil-works

    Local Linux micro-VMs with a TypeScript control plane for programmable network, filesystem, and secret injection policies.

    Use caseUse it to wrap agent tool calls in disposable QEMU or krun VMs while keeping credentials and egress on the host.

    microvmsandboxtypescriptcontrol-plane
  2. 02harnessGitHub
    Mecatl↗

    stacklok

    Cloud-native open-source agent harness with provider-agnostic loop, durable state, permissions, gRPC/HTTP APIs, and optional local microVM execution.

    Use caseUse it to run the same agent loop locally or on Kubernetes with Redis-backed sessions and deny-dominant permissions.

    harnesskubernetescloud-nativepermissions
  3. 03repoGitHub
    zerocore-ai/microsandbox↗

    zerocore-ai

    Open-source local-first hardware-isolated sandboxes with instant startup and OCI compatibility for agent workloads.

    Use caseUse it as an embeddable microVM backbone for untrusted code execution in agent pipelines.

    microvmlocaloci
  4. 04toolOrigins AI / product
    Prime Sandboxes↗

    Prime Intellect

    Hardware-isolated Linux microVMs for agentic RL at scale with large immutable environment catalog and high concurrency.

    Use caseUse it to train or evaluate agents against reproducible microVM environments at thousands of concurrent instances.

    microvmrlevalcatalog
  5. 05harnessnpm
    @ai-sdk/harness↗

    Vercel AI SDK

    Experimental HarnessAgent plus sandbox session interfaces for network-aware agent loops in the AI SDK.

    Use caseUse it to bootstrap typed harness sessions with Vercel or E2B sandboxes and schema-backed outputs.

    sdkharnesstypescriptsandbox
  6. 06toolGitHub Marketplace
    Agent Security Harness↗

    msaleme

    CI action that runs protocol-level adversarial tests against MCP/A2A endpoints with PASS/FAIL/INCONCLUSIVE verdicts.

    Use caseUse it as a gate to probe live agent endpoints for policy violations without false PASS on unreachable targets.

    securityevalcimcp
  7. 07repoGitHub
    e2b-dev/ai-sdk-sandbox↗

    e2b-dev

    E2B sandbox backend for Vercel AI SDK so HarnessAgent code runs inside microVMs.

    Use caseUse it to wire E2B sessions into AI SDK tools and harnesses with restricted surfaces.

    e2bai-sdksandbox
  8. 08repoGitHub
    RyanAlberts comparisons↗

    RyanAlberts

    Category-ranked top harnesses regenerated weekly from live star and capability data.

    Use caseUse it to shortlist progressive-disclosure, coding, or research harnesses by current metrics.

    rankingcomparisonharness

02 — In the wild

Articles and releases

  1. releaseAWS Open Source Blog
    AWS launches Strands Box open-source agent sandbox↗

    Fernando Dingler / AWS

    Strands Box combines OS isolation with Dogwood policies, Strands Shell, and Monty so agents face fine-grained action controls rather than YOLO mode. It is Apache-2.0, macOS-first, and works with any harness.

    awssandboxreleasepolicy

    Shipped

  2. articleCNCF
    Mecatl: the case for a cloud-native agent harness↗

    Craig McLuckie

    Stacklok argues agent runtimes must be distributed systems with durable state, leases, and replaceable execution planes. Mecatl ships as the open reference that keeps the loop independent of clients and infrastructure.

    cncfharnesskubernetesessay

    Read

Get the daily tools digest

RSS