آموزش DeepSeek Harness: ساخت ایجنت‌های برنامه‌نویسی با مدل‌های رایگان

عنوان دوره : Udemy – DeepSeek Harness: Build AI Coding Agents with Free Models

سطح : مبتدی

زمان کل دوره : 2 ساعت و 12 دقیقه

تعداد فايل ها : 54 (1080p)

سال آموزش : 2026

فایل تمرینی : دارد: HTML (15)، PDF (7)، ZIP (4)

مربی دوره : Dr. Amar Massoud

نرم افزار : –

زیرنویس انگلیسی : دارد

Udemy – DeepSeek Harness: Build AI Coding Agents with Free Models

This course contains the use of artificial intelligence.

DeepSeek Harness is an open-source, MIT-licensed agent harness built on a single idea: everything is a plugin. Models, tools, skills, sessions, the agent loop itself — all swappable. That makes it an unusually good place to learn how coding agents actually work, because nothing is hidden behind a product.

The catch with most agent courses is the bill. This one has none. The harness runs locally on an ordinary Windows, macOS or Linux machine, and the thinking happens on a free cloud model — Gemini’s free API tier as the primary route, with explicit free OpenRouter models as a second provider for when a quota runs out. No paid API key. No GPU. No subscription. There is an optional section for running a small model entirely locally through Ollama, for work you cannot send anywhere.

Anyone can film an agent succeeding. The hard part — the part that decides whether you can trust one of these things at work — is telling the difference between an agent that did the job and an agent that says it did. That distinction runs through the whole course.

You will read trajectories: the actual model requests, tool arguments, tool responses and commands recorded for every session. You will set permission modes so an exploration task cannot write a file, rather than politely asking it not to. You will review diffs before approving them, separate pre-existing test failures from ones you just caused, and require the agent to show real test output instead of a summary.

The capstone has an automatic-fail rule that says everything about the course’s priorities: if your agent claims success but the trajectory shows no test was ever run, you fail — even if the code happens to be correct.

Everything runs against a small practice repository built for this course: zero dependencies, sixteen tests, and defects that genuinely reproduce.

Configure a custom model provider from scratch — base URL, protocol, credential, model catalogue — and swap providers without touching your project. Write persistent project instructions the agent actually follows. Package a reusable skill, a slash command and a tool as plugins. Run the harness headless for automation, and drive it from Python. Decide when a second agent helps and when it just duplicates work.

DeepSeek Harness is a developer preview. It ships breaking changes. The course names the exact version everything was recorded against, ships a version sheet, and — more usefully — teaches you the methods for finding the current answer yourself rather than memorising strings that expire. Free tiers are a commercial condition, not a property of the software, and the course says so plainly instead of pretending otherwise.

You should be able to open a terminal and read a small JavaScript file. You do not need any prior AI or agent experience.

This course is independent and is not affiliated with, endorsed by, or sponsored by DeepSeek, Google, OpenRouter or Ollama.

What you’ll learn:
– Install and run DeepSeek Harness locally on Windows, macOS or Linux with no GPU and no paid API key
– Explain precisely what a harness adds to a language model, and why a chatbot cannot do this work
– Configure Gemini’s free tier as a custom provider — base URL, protocol, credential and model catalogue
– Add explicit free OpenRouter models as a second provider, and switch between them without touching your project
– Find today’s free models that actually support tool calling, instead of relying on a list that expires
– Give an agent controlled access to files and the terminal using read-only, workspace-write and full-access modes
– Use plan mode and approval review to stop an agent before it does something you did not ask for
– Read a session trajectory and tell the difference between what an agent claims and what it can prove
– Diagnose the failure that stops most people: a correct key and URL where every request is still refused
– Write persistent project instructions, reusable skills and custom slash commands the agent follows
– Build and package a custom tool as a plugin, and reuse it in a second project unmodified
– Run the harness headless for automation and drive it programmatically from Python
– Decide when multiple agents genuinely help, and when they just duplicate work and burn context
– Deliver a complete repository-maintenance agent whose work is backed by real test evidence
– Optionally run a small model fully locally through Ollama, and judge honestly when it is not good enough

Requirements:
– An ordinary Windows, macOS or Linux computer — no GPU required
– Node.js ^22.19 or 24 and up (the course shows you how to check and install it)
– Comfortable opening a terminal and running a command
– Able to read a small JavaScript file well enough to recognise a bug when it is pointed out
– A Google account for the free Gemini API tier, and an OpenRouter account — both free to create
– No paid AI subscription, no API credit, and no prior AI or agent experience required

Who this course is for:
– Developers who have hit the paywall on AI coding tools and want to keep learning anyway
– Junior and mid-level engineers who want to understand how coding agents work rather than just using one
– Students and self-taught developers who need a zero-cost path to real agent experience
– Developers in regions where paid API access is impractical or unavailable
– Platform and DevOps engineers evaluating agent harnesses before committing a team to one
– Technical leads who need to write sensible policy about what an AI agent is allowed to touch
– Anyone who has watched an agent confidently report success it did not earn, and wants to catch it next time