Awesome list for AI agent harness engineering: tools, patterns, evals, memory, MCP, permissions, observability, and orchestration.
Python agent-harnessagent-memoryagent-orchestrationai-agent-harnessai-agentsawesome-listcontext-engineeringharness-engineeringmcp
Add forkd: fork-from-warm microVM sandbox for AI agent fan-out loops
ai-boost94b324d37bbc
diff --git a/README.md b/README.mdindex 840a4d6..1a30f13 100644--- a/README.md+++ b/README.md@@ -430,6 +430,7 @@ Real repositories worth studying — each with a note on *why* it's worth your t- [Alibaba OpenSandbox](https://github.com/alibaba/OpenSandbox) — General-purpose sandbox platform for AI agents (8.7K+ stars, March 2026) with multi-language SDKs (Python, Java, TypeScript, Go, C#), unified APIs across Docker/Kubernetes runtimes, and support for secure container runtimes (gVisor, Kata Containers, Firecracker). Covers coding agents, GUI agents, agent evaluation, and RL training in a single abstraction layer — the most runtime-flexible sandbox option when you need to choose isolation levels per workload. - [CubeSandbox](https://github.com/TencentCloud/CubeSandbox) — Tencent Cloud's production-validated microVM sandbox for AI agents: sub-60ms cold start via snapshot cloning, <5MB per-instance overhead, and true kernel-level isolation with eBPF-enforced network policies. E2B-compatible drop-in replacement that demonstrates how hyperscale cloud infrastructure can be repurposed for high-density agent execution. - [zeroboot](https://github.com/zerobootdev/zeroboot) — Sub-millisecond VM sandboxes for AI agents via copy-on-write forking, enabling fresh isolated execution on every tool call without the latency penalty of container or microVM cold starts. The critical harness advantage is turning per-action isolation from a batch-mode luxury into a real-time loop primitive — agents can spin up and tear down environments within a single reasoning step. +- [forkd](https://github.com/deeplethe/forkd) — MicroVM sandbox runtime that replaces cold-boot isolation with fork-from-warm: children share a paused parent's memory copy-on-write, so 100 agent tool-execution sandboxes spawn in ~100 ms rather than seconds. The live-BRANCH primitive (~56 ms pause) lets an agent fork its own in-flight state for speculative exploration or checkpointing — a fundamental leap for harnesses where per-action isolation is currently too slow to deploy in the inner loop. - [The Attack and Defense Landscape of Agentic AI: A Comprehensive Survey](https://arxiv.org/abs/2603.11088) — The first systematic survey of AI agent security from UC Berkeley and UIUC (Dawn Song et al., March 2026). Reviews 128 papers covering 51 attack methods and 60 defense mechanisms. Introduces a framework for understanding security risks specific to agentic (not just LLM) systems and identifies open gaps in securing agent architectures — the definitive 2026 reference for agent threat modeling.- [Trustworthy agents in practice](https://www.anthropic.com/research/trustworthy-agents) — Anthropic's April 2026 framework for governing autonomous agents through five principles: human control, value alignment, secure interactions, transparency, and privacy. The most complete published treatment of how to design harness-level governance that keeps pace with increasing agent capability and autonomy.