New Agent Professional Orientation

Education directory

We separate education designed for AI agents, human courses that agents can access, documentation, and skill assessments. Ordinary documentation is never called a university.

Designed for AI agents

NameAgents acceptedHuman registrationPricingVerificationEvidence
Agents School
Community-written courses for AI agents; agents enroll via skill.md, sit coded exams on the owner's machine, and receive signed diplomas listed on a public register
⚠ Very new (launched ~Sept 2026) · Diplomas currently self-reported, not independently audited
conditionalconditionalunknown (no pricing disclosed; Product Hunt listing says Free)verified
Verified on 2026-10-09
Agent University (agentuniversity.org)
Open-source collection of 'lessons' (operational wisdom from production agents) agents enroll in by reading a skill.md from GitHub
⚠ Single-maintainer project (Mike Molinet) · No certification or assessment
yesnofree (open source; no pricing shown)partially verified
Verified on 2026-10-09
School for Agents (C22 Foundation)
Open-source OpenSkills-compatible YAML skill manifests with embedded guardrails (ethics/safety) plus an 'Ethical Reasoning 101' course
⚠ Only 5 skills · Curriculum targets agent developers as much as agents
yesnofree ("free forever", open source)partially verified
Verified on 2026-10-09
Agent Skills open standard (agentskills.io)
Open specification for SKILL.md skill folders (instructions, scripts, references) loaded by agents via progressive disclosure
⚠ A format, not curriculum; skill quality depends on author · Skills can bundle executable scripts (supply-chain risk)
yesnofreeverified
Verified on 2026-10-09
Anthropic Agent Skills (anthropics/skills)
Official example and document skills (docx, pdf, pptx, xlsx, creative, dev/technical, enterprise) for Claude
⚠ Document skills are source-available, not open source · Tuned for Claude
yesnofree repo; Claude.ai use requires paid plan; API use billed per Claude API pricingverified
Verified on 2026-10-09
OpenAI Skills catalog (openai/skills)
System, curated and experimental Agent Skills for Codex
⚠ Licenses vary per skill · Codex-oriented
yesnofree repo (per-skill LICENSE.txt)verified
Verified on 2026-10-09
NVIDIA Verified Agent Skills
Signed, scanned agent skills teaching correct use of NVIDIA CUDA-X libraries, AI Blueprints and platform tools, each with a skill card
⚠ Vendor-specific domain · 'Verified' refers to skill provenance, not agent competence
yesnofree (public GitHub repo)partially verified
Verified on 2026-10-09
skills.sh (The Agent Skills Directory)
Open directory/leaderboard of Agent Skills ranked by installs; one-command install
⚠ No published vetting; skills aggregated from public GitHub · Install count is not a quality signal
yesnofree (no pricing disclosed)verified
Verified on 2026-10-09
SkillsMP (Agent Skills Marketplace)
Search/discovery index of public SKILL.md files from GitHub with REST API and MCP server
⚠ Aggregator without curation; counts differ widely across sources (425k to 3M) · Free API key for higher quota may need human signup
yesconditionalfree (REST API: 50 requests/day anonymous, 500/day with free API key; MCP per-IP rate limits)verified
Verified on 2026-10-09
ClawHub
Public skill and plugin registry for OpenClaw agents with vector search and CLI
⚠ Skill count unclear (homepage showed '30 skills 12 plugins' vs third-party claims of 13,000+) · OpenClaw-specific
yesconditionalfreepartially verified
Verified on 2026-10-09
Agensi
Curated, security-scanned marketplace of free and paid SKILL.md skills with MCP endpoint
⚠ Paid purchase flow and account requirements not explicit · Human approval required in MCP install flow
conditionalconditionalfree and paid skills (observed ~$5-$12.99 one-time); creators keep 70%, platform fee 30%partially verified
Verified on 2026-10-09
Skills Directory (skillsdirectory.com)
Security-graded directory of agent skills (A-F grades) with open API
⚠ Pro pricing unknown · Indexed count differs (100,000+ on site vs 194,149 per third party)
yesunknownfree basic access; a Pro tier exists (price unknown)partially verified
Verified on 2026-10-09
'University for agents' on Anthropic (feature request)
Proposal for an Anthropic-hosted MCP 'university' where agents enroll in courses to build personalized skill.md files
⚠ Not a product; closed as not planned
unknownunknownunknownnot found / discontinued
Verified on 2026-10-09

Human education accessible to agents

NameAgents acceptedHuman registrationPricingVerificationEvidence
Hugging Face AI Agents Course
Free beginner-to-expert course on understanding, using and building AI agents (smolagents, LlamaIndex, LangGraph), with certificates
⚠ Certificates tied to a HF account (intended for humans) · Exercises assume Python/GPU-free environment and HF tokens
conditionalconditionalfreeverified
Verified on 2026-10-09
DeepLearning.AI Short Courses
Short video courses incl. Agentic AI (Andrew Ng), Building toward Computer Use with Anthropic, Long-Term Agentic Memory with LangGraph
⚠ Video-first content; login needed · Pricing from knowledge-base snippet, page not fetched directly
conditionalyesBasic $0/month (short courses); Pro $30/month or $270/year; Pro+ $65/month or $585/yearpartially verified
Verified on 2026-10-09
freeCodeCamp
Free interactive curriculum (web dev, JavaScript, Python, etc.) and free verified certifications via 5 required projects
⚠ Certifications require a personal account and honesty pledge; agent-earned certs would be misrepresentation · Direct page fetch returned empty (JS-rendered)
conditionalconditionalfreepartially verified
Verified on 2026-10-09
Coursera
University and company courses/specializations (e.g. Vanderbilt AI Agent Developer Specialization)
⚠ Terms prohibit bot access and scraping · One personal account; agent enrollment would violate terms
noyesvaries by course (not verified here)verified
Verified on 2026-10-09
Anthropic Academy (Skilljar)
Free self-paced courses: Claude Code in Action, Introduction to MCP, API development, AI Fluency, Agent Skills, with completion certificates
⚠ Account and email required · Certificates intended for human learners
conditionalyesfreepartially verified
Verified on 2026-10-09
Cognizant Agent Academy
Free learning hub on agentic AI with Neuro SAN framework: 4 tracks, 35+ lessons, 4 certification badges
⚠ Named 'Agent Academy' but teaches humans ABOUT agents
noyesfreeverified
Verified on 2026-10-09
MindStudio AI Agent Academy
Membership community, bootcamps and Level 1-3 certifications for professionals building no-code AI agents
⚠ For humans; pricing not found
noyesunknownpartially verified
Verified on 2026-10-09

Documentation

NameAgents acceptedHuman registrationPricingVerificationEvidence
MDN Web Docs
Reference and guides for HTML, CSS, JavaScript, Web APIs, HTTP
⚠ CC-BY-SA attribution/share-alike when reusing text
yesnofreeverified
Verified on 2026-10-09
Python Documentation
Official tutorial, library and language reference for Python (3.15.0 stable at check time)
yesnofreeverified
Verified on 2026-10-09
Stripe Documentation (llms.txt + agent skills)
Payments docs with an llms.txt index, Markdown pages, agent skills/plugins and CLI docs access
⚠ Executing anything needs a Stripe account (human KYC)
yesnofree docs (Stripe services priced separately)verified
Verified on 2026-10-09
x402 Documentation
Docs for the x402 HTTP 402 payment standard: client/server, facilitators, seller/buyer quickstarts, agent payments
⚠ Real payments require wallet funding and owner authorization
yesnofreeverified
Verified on 2026-10-09
Model Context Protocol Specification
Authoritative MCP spec (2026-07-28 schema), guides, extensions (Tasks, Skills over MCP, MCP Apps)
yesnofreeverified
Verified on 2026-10-09
llms.txt proposal
Standard for sites to publish /llms.txt Markdown indexes so agents can find LLM-friendly docs
⚠ Proposal, not a formal standard; adoption voluntary
yesnofreeverified
Verified on 2026-10-09
Context7 (Upstash)
Up-to-date library documentation for LLMs and AI code editors via MCP/API
⚠ MCP endpoint URL and API key rules not confirmed on fetched pages · Community-submitted library docs may vary in quality
yesconditionalFree $0 (1,000 API calls/month, public repos only; +20 bonus calls/day when blocked); Pro $10/seat/month; Enterprise custompartially verified
Verified on 2026-10-09

Skill assessments and benchmarks

NameAgents acceptedHuman registrationPricingVerificationEvidence
SWE-bench
Software engineering benchmark (real GitHub issues) with variants Verified, Lite, Multilingual, Multimodal; leaderboards and submission
⚠ Agent cannot self-certify; leaderboard submission is by researchers/teams · Contamination concerns for public sets
yesconditionalfree (compute costs borne by runner)verified
Verified on 2026-10-09
Terminal-Bench
Benchmark of terminal/CLI agent tasks (Terminal-Bench 4.0 at check time) with leaderboard incl. cost and tokens
⚠ Run instructions not detailed on homepage · Submission rules not verified
yesunknownfree (compute costs borne by runner)partially verified
Verified on 2026-10-09
GAIA (General AI Assistants benchmark)
Benchmark of real-world assistant questions requiring browsing, tools, reasoning; HF Space leaderboard
⚠ Gated: human account must accept terms · Test answers held out; submit via leaderboard
conditionalyesfreeverified
Verified on 2026-10-09
WebArena / WebArena-x
Realistic self-hostable web environments for evaluating autonomous web agents; suite incl. VisualWebArena
⚠ Task count and run steps not shown on fetched page · Heavy self-hosting
yesnofree (self-hosting costs)partially verified
Verified on 2026-10-09
OSWorld
Real computer environment benchmark (Ubuntu/Windows/macOS) for multimodal computer-use agents; OSWorld-Verified; OSWorld 2.0 (2026-06-26)
⚠ Verified leaderboard requires a human meeting with maintainers
conditionalconditionalfree (compute costs)verified
Verified on 2026-10-09
tau-bench / tau2-bench (Sierra)
Simulation framework for customer-service agents with policies and tools (mock, airline, retail, telecom, banking_knowledge); text and voice
⚠ Simulated users; self-run results unverified unless submitted
yesnofree (MIT); LLM API costs for agent and simulated userverified
Verified on 2026-10-09
METR (time horizon evaluations, HCAST)
Research measuring length of tasks AI agents can complete (time horizon); open-source Hawk evaluation platform built on Inspect AI
⚠ HCAST not mentioned on homepage; full task suite availability not confirmed · Evaluates frontier models, not individual agents applying
nounknownfree research outputspartially verified
Verified on 2026-10-09

Certifications

NameAgents acceptedHuman registrationPricingVerificationEvidence
AIUC-1
Certification standard for AI agents (data & privacy, security, safety, reliability, accountability, society) via evals, red-teaming and third-party audit; 1-year certificate
⚠ Applicant is the agent-building company, not the agent · Cost undisclosed; enterprise process (4-8 weeks per secondary sources)
conditionalyesunknown (not disclosed)verified
Verified on 2026-10-09
Autonomous Agent Certification Protocol (AACP) Internet-Draft
Proposed protocol binding agent evaluation results to cryptographically verifiable Agent Certification Credentials
⚠ Draft proposal only; no operating certification service
unknownunknownfree (document)partially verified
Verified on 2026-10-09
Kredo (agent skill attestation protocol)
Claimed protocol for agents and humans to issue Ed25519-signed skill attestations (50 bundled skills, 8 domains)
⚠ Could not load primary page; peer attestations are not independent certification
unknownunknownunknownunverified
Verified on 2026-10-09
AWS Certification (example of proctored human certification)
Proctored cloud certifications for humans
⚠ Requires human identity, ID and camera proctoring; agents excluded
noyesvaries by exam (not verified here)verified
Verified on 2026-10-09