deer-workflow
An open-source Dynamic Workflow runtime that combines deterministic TypeScript orchestration with replaceable Agent runtimes.
它通过自动扫描 SDK 表面并使用真实客户端进行一致性测试来衡量 API,不像传统 mock 服务器那样盲目响应;另外还可通过 Incus 提供真实的虚拟机,实现网络仿真。
One binary. One port. No account, no bill.
AWS has LocalStack. Azure has Azurite. European clouds have had nothing, and their users test against a paying account or not at all.
feint emulates their APIs so that SDKs, CLIs and Terraform run against your laptop. Scaleway, Outscale and Exoscale are the first three — the architecture exists so a fourth changes nothing outside its own pack, and the scan that keeps them honest reads whatever SDK or API description that provider publishes.
The name is the fencing move: feinte, from Old French feindre — a movement made to look like the real one, so the opponent commits. That is the whole test here. The official client commits, and cannot tell.
The mark draws the same argument: a solid line for the surface actually served, a dashed one leaving the same corner at the same angle for what the provider announced and nobody has driven yet. docs/brand.md has the files and what may be done with them.
feint serve
# feint dev listening on 127.0.0.1:4599
# scaleway 55 routes
# outscale 20 routes
# exoscale 14 routes
# machines none
That is one endpoint for three clouds. They do not collide in URL space —
Scaleway serves //v/…, Outscale POST /api/v1/, Exoscale
/v2/ — so one http.ServeMux hosts all three, and the server refuses
to start if two packs claim the same route.
Every command in that recording really ran: the emulator starts, the official
scw CLI creates a server, the official Terraform provider applies a
configuration through api_url, and the emulator stops. It is generated from
tools/demo/quickstart.tape with mise run demo,
so a command that breaks breaks the demo. A recording nobody can re-run is a
claim, and the whole argument here is that a claim is not a proof.
A second tape covers the part
Local emulator for European cloud APIs — Scaleway, Outscale, Exoscale. One binary, one port, no account.
根据分类、Topic 和编程语言匹配的相似项目。
An open-source Dynamic Workflow runtime that combines deterministic TypeScript orchestration with replaceable Agent runtimes.

TokHub is an open-source monitoring, recommendation, and OpenAI-compatible gateway system for AI API intermediaries, combining public status pages, supplier rankings, user workspaces, layered probing, metering, alerts, encrypted keys, and Docker self-hosting.

A comprehensive step-by-step tutorial series that teaches AI agents from scratch, covering function calling, agent loops, ReAct, reflection, plan-and-execute, multi-agent systems, orchestration, evaluation, and related concepts like RAG, GraphRAG, and LangChain.