27 free resources on this topic. Everything here is free and hand-picked. You can also search within this topic.
A DeepLearning.AI course built with Prior Labs, about 1.5 hours at the intermediate level, on using tabular foundation models such as TabPFN for prediction on spreadsheet-style data and connecting them to data science agents. Enrollment requires a free account.
An open GitHub repository from Jane Street with agent skills for AI coding assistants, including a Socratic tutor skill, published with a license and contribution guide.
A Columbia University-hosted benchmark that tests AI agents on reverse engineering complex binary software. The page reports results from evaluations run with OpenAI, stating that GPT-6 Astra solves nearly all tasks while other evaluated models stay below 60%, and describes pilot evaluations of harder settings.
A Harvard SEAS graduate course on modern AI systems, covering agents and system optimizations. Course materials are publicly available on the course site.
Paper by Stephen Omohundro (2008) arguing that sufficiently advanced goal-directed AI systems would tend to develop basic drives such as self-preservation, resource acquisition, and goal preservation, and discussing what those drives imply for designing safe AI. Hosted on the Self-Aware Systems site.
A free live hour with Yash Thakker covering the AI safety concepts every builder and operator needs: the core failure modes to know before hitting them, a live look at AgentBeam catching an unsafe agent trajectory, and a checklist of best practices for running agents safely. Free with email registration; a recording is included.
The personal site of Noah Shinn, a programmer and researcher who was one of the first employees and a researcher at Sierra. His papers include tau-bench, a benchmark for tool-agent-user interaction (ICLR 2025), work on code-editing instruction following, type prediction, and the Reflexion paper on language agents with verbal reinforcement learning.
A step-by-step playbook from Late Checkout for building an AI-native organization, covering agent skills, MCP, knowledge layers, hiring for judgment, and what replaces Agile. Published April 2026.
A free self-paced curriculum for the Forward Deployed Engineer role — the person who sits with a customer and turns an AI model into a system that actually runs in production. 11 modules, 150+ topic articles, 700+ open-book questions and 15 daily drill questions, covering systems thinking, problem decomposition, LLM/RAG/agent architecture, data and event systems, physical AI and robotics, plus client discovery, integration, deployment and stakeholder management. It also publishes a map of 15 layers of the modern AI engineering stack drawn from real FDE and AI-engineer job specs, with links out to where to learn each one.
Why I recommend it: The whole curriculum opens without an account or a card — I checked — and an account only saves your progress. There is no pricing page anywhere on it, so nothing is being held back behind a paywall today. Worth knowing why this matters: 'Forward Deployed Engineer' is a job title that is actually being hired for right now, and it pays like engineering while rewarding the consulting side most engineers avoid. Two flags. It is one person's independent site, not an accredited course or a certificate anyone will recognize — the value is the material, not a credential. And it makes a strong claim, that foundation models are becoming a commodity and the money lives in the last mile; that is an argument, not a measured finding, so read it as a well-argued position.
Amazon's agentic coding environment: you write a spec, and its agents plan, build and check code across a codebase. Runs as a desktop app.
From the site: Kiro helps developers and teams do their best work: turn prompts into executable specs, validate code correctness to find bugs unit tests miss, and build across large codebases with parallel agents that learn from every session.
Why I recommend it: Freemium. The free plan gives you 50 credits a month plus access to Claude Sonnet 4.5 and open-weight models, subject to rate limits; paid plans start at $20 a month. Fine for trying agentic coding, not for all-day use.
An open-source platform for running coding agents such as Claude Code and Codex across your own machines and specialising them for domains like CAD, PCBs, robotics and games, with optional open hardware.
From the site: Follow your curiosity. Build across disciplines. Open-source software and hardware for polymaths in the making. - autonomous-ai/openharness
Why I recommend it: The software is free and open source. The agents you run inside it may cost money, and the companion hardware is a separate purchase — the repo itself is the free part.
An open-source platform for building and running AI apps and agents, with a hosted option and a full self-hosted Docker install.
From the site: Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without ...
Why I recommend it: More structure than a chat window: prompts, data and logs kept in one place. Start with their Docker install and one small app.
An open-source visual builder for AI workflows and agents — drag components together, then export the flow as an API. MIT licensed.
From the site: Langflow is a powerful tool for building and deploying AI-powered agents and workflows. - langflow-ai/langflow
Why I recommend it: Useful for seeing what an AI 'agent' actually is: a chain of steps you can look at, not magic.
A free arXiv preprint on recursive self-improvement in AI agents trained inside evolving simulated worlds.
Why I recommend it: Technical, and central to the safety debate about systems that improve themselves.
A terminal-based coding agent you run locally to read, write and refactor code from the command line.
From the site: A terminal-based coding agent
Why I recommend it: If you already live in a terminal, this is a lighter way to try agentic coding than a full IDE.
A company building governed, verifiable agentic infrastructure for enterprise systems, with free research and publications on its site.
From the site: Emergence builds mission-critical agentic infrastructure for enterprise. Verified, governed AI agents that plan, reason, and act across the most complex systems.
Why I recommend it: Their research and reports are free to read; the platform itself is an enterprise product, so treat the writing as the resource here.
An open format for telling coding agents how to work in your repository, now used by tens of thousands of open-source projects.
From the site: AGENTS.md is a simple, open format for guiding coding agents. Think of it as a README for agents.
Why I recommend it: If you're experimenting with AI coding tools, this is the convention to follow so your instructions actually get read.
A free report from Emergence AI surveying agentic AI in enterprise settings — where agents are being deployed and what governs them.
Why I recommend it: Read it as an industry vendor's view of where agent work is going, useful context if you're targeting AI roles.
A self-hosted, MIT-licensed AI agent with persistent memory that builds skills over time and reaches you on Telegram, Discord and other channels.
From the site: Self-hosted AI agent that remembers your projects, builds skills automatically, and reaches you on Telegram, Discord & more. MIT license. No tracking.
Why I recommend it: Free and open source, and it runs on your own machine — worth a look if you don't want your project context sitting on someone else's server.
The running record of what changed in each Polytoken release, alongside its free documentation and quickstart.
Why I recommend it: Handy if you use the tool and want to know what broke or improved.
A free local-first AI coding agent that runs as a daemon on your own machine and executes tools against your development environment.
Why I recommend it: Free to use and runs locally, so it is a low-risk way to try agentic coding.
A free, open-source security scanner that checks AI agent skills and MCP servers for prompt injection, data exfiltration and supply-chain risks before you install them.
Why I recommend it: If you install agent skills, scan them first. This is the free tool to do it with.
DeepSeek's free open-source agent harness, built on an "everything is a plugin" architecture, with a local web interface you can run in one command.
Why I recommend it: Worth a look if you want to run AI agents locally without paying for a hosted platform.
A free Claude Code skill where two AI models harden a build plan before any code is written, then swap roles so whoever built it never grades it.
Why I recommend it: A good habit to borrow even outside code: have something other than the author check the plan.
Free online session with Brandon Waselnuk of Unblocked on giving AI coding agents real context so they produce usable work the first time, including a live before-and-after demo and two open-source tools to try.
Details: Free, online, 9:00 AM PT on September 23, 2026. Registration required.
Official documentation for ChatGPT computer use, explaining what the agent can do and its safety limits.
Why I recommend it: Read the limits section carefully before you hand an agent anything connected to your accounts.
Developer-focused resource for building faster with AI coding tools and agent workflows.
Why I recommend it: Useful if you are building a technical portfolio — pair AI-assisted builds with a clear write-up of what you decided and why.
