Free AI agent tools, checked by hand and by robot
Every free LLM API, MCP server, agent framework and free-tier service you need to build an agent in 2026 — with the free limit, the rate limit, whether a credit card is required, and the date a human last confirmed it.
free-ai-agent-stack is a hand-verified catalogue of 384 free resources for building AI agents in 2026, across 5 categories and 222 that need no credit card. Every entry records its free limit and the date a human last checked it. Last export: 2026-09-24.
- resources catalogued
- 384 resources catalogued
- need no credit card
- 222 need no credit card
- verified within 30 days
- 100% verified within 30 days
Start here
The entries most people should look at first.
Serverless inference that runs inside a Worker — no egress cost, no cold-start management, free daily budget.
llm-apis✅ no card
Version-pinned, up-to-date documentation for thousands of libraries — the single highest-value server for coding agents.
mcp-servers
Terminal pair-programmer that commits every change to git and maps your repo into the model's context.
agent-tools✅ no card
TypeScript-first authentication library that lives inside your app — no third party in the request path.
free-tiers✅ no card
Anthropic's official skills repository — first-party packs that teach Claude Code and other agents house workflows: document handling, spreadsheet and slide authoring, PDF work and more.
skills
Frontier and open models behind a GitHub PAT — the lowest-friction free tier if you already have a GitHub account.
llm-apis✅ no card
Drives a real Chrome instance for performance traces, console errors and network inspection.
mcp-servers
Autonomous coding agent for VS Code, JetBrains and the terminal, with human approval at every step.
agent-tools✅ no card
Browse all categories
- Free LLM APIs70 entries · 58 without a card
A free LLM API is the starting point for almost every agent project, and it is the category where lists rot fastest: Cerebras removed its no-card free tier in July 2026, Google moved its Pro models behind billing in April, and Half the providers quoted in older guides now require a payment method before the first request..
- MCP Servers64 entries
Model Context Protocol servers are how an agent reaches the outside world — files, databases, browsers, tickets, payments. The official registry lists what exists; this page tells you what each server actually needs before you run it: the install command, the transport, and whether it wants a token, a connection string or nothing at all..
- Agent Tools82 entries · 82 without a card
“Free plan available” is the least useful sentence in developer marketing. This page puts the actual number next to each tool: Gemini CLI's 1,000 free requests per day, GitHub Copilot Free's 2,000 completions, Windsurf's unlimited tab completion with a quota-limited agent..
- Free Tiers90 entries · 82 without a card
An agent needs more than a model: somewhere to run, somewhere to remember, something to search, and something that tells you when it breaks. This is the full stack, with the free limit and a cold-start-risk rating for each service — because free hosting that sleeps for sixty seconds behaves very differently from free hosting that does not..
- Agent Skills78 entries
An agent skill is a folder of instructions and scripts that teaches an agent a workflow it was not trained on: how to render and check a PDF, how to deploy to Vercel, how to repair a failing CI job. Unlike an MCP server, a skill adds no moving part — no process to run, no token to hold, no network service to keep alive.
Building an agent, not a human?
The whole dataset is published as JSON and an llms.txt summary, with stable IDs and verification dates, so an agent can cite it without scraping HTML.
How verification works
- A maintainer opens the vendor's pricing page and records the free limit, the rate limit and whether a card is required.
- The date and their handle go into the entry — every card on this site shows so you can judge how current the claim is.
- A robot re-checks every URL daily and flags dead links after three consecutive failures.
- Anything unverified for 30 days is thrown back into the weekly triage issue.
Who maintains this
Built and maintained by Adarsh Kushwah at Dev Animecx. Every entry names the handle of whoever verified it; the catalogue is independent and not affiliated with any listed vendor.
Recently re-verified: Notion API · A11y Audit · Alireza Rezvani Claude Skills · Figma Agent Skills · Exa Contents
Questions people ask
What is free-ai-agent-stack?
A hand-verified catalogue of 384 free resources for building AI agents in 2026 — LLM APIs, MCP servers, agent tools, infrastructure free tiers and agent skills — across 5 categories. Every entry records its free limit, whether a credit card is required, and the date a human last verified it.
How often is this data updated?
Every URL is re-checked daily by a robot, which flags dead links after three consecutive failures. Human verification dates are recorded per entry and anything older than 30 days is returned to the weekly triage list. The dataset was last exported on 2026-09-24.
Can I use this data in my own project?
Yes. The data is licensed CC BY 4.0 and the code is MIT, so you may republish and adapt it commercially provided you attribute it. The full catalogue is published as JSON and as an llms.txt summary with stable IDs, so you can consume it directly rather than scraping the pages.
How do I know a claim here is accurate?
Every entry names the date and handle of whoever verified it, and the catalogue keeps resources that quietly stopped being free — flagged rather than deleted — so the record shows where claims failed. Entries that require a credit card are labelled, never presented as free.
Is there an API for this catalogue?
There is no rate-limited API because none is needed: the whole dataset is static JSON at /data/all.json, with per-category files alongside it, plus /llms.txt and /llms-full.txt for language models. The files are regenerated from the YAML source on every change and are cacheable indefinitely.