Build for AI Agents: LLM SkillHub Challenge 2026

Build for AI Agents: LLM SkillHub Student Challenge 2026 AI is moving from chatbots to AI agents that actually get work done. But agents still waste tokens, lose track of multi-step workflows, and need the same instructions again and again. This challenge asks you to fix that by teaching an agent a skill. The Challenge Build a skill: a reusable set of instructions (a SKILL.md file, plus optional scripts and reference files) that teaches an AI agent to do one specific task better, faster and more reliably. Skills work with agents like Claude Code, Cursor, Codex, Gemini CLI and opencode. Your skill can help with anything: coding, research, education, productivity, data analysis, career guidance, automation, or a problem specific to India. You don't need to build an app or train a model. Teach an existing agent something useful. What You'll Do Pick a real-world problem, ideally one you have seen up close. Build an AI-agent skill that solves it. Publish it as a PUBLIC skill on LLM SkillHub (llmskillhub.com). (Optional, good to have) Create evaluation cases that define what "good" looks like, and run them. Test your skill with real users and improve it based on their feedback. Submit your entry here on Unstop. What to Submit The link to your public skill on LLM SkillHub A short write-up (up to 500 words): the problem, who has it, how your skill solves it, and what you changed after testing User feedback: how many people tried it and what they said Optional, but good to have: your eval and its results, and a demo video of up to 2 minutes Important Your skill must be public. Skills kept in a private space can't be judged. Provide the email ID linked to your LLM SkillHub account on the Unstop registration form (you sign in to LLM SkillHub with Google or GitHub). Publish early: every public skill is reviewed by a person before it is listed. Each participant can submit as many as skills they feel like, no restrictions on number of skills submitted. Judging Criteria Real-world usefulness Quality and reliability of the skill Problem-solving and originality Documentation and usability Feedback from real users Evaluation performance (if you include an eval) What You Get A certificate for everyone with a valid submission Top 10: T-shirt + Sipper + Keychain Ranks 11–20: T-shirt Mentorship from industry experts for the best skills that target a real problem and solve it Preference for future internship opportunities for winners (a preference, not a guaranteed selection) Your skill published and showcased on LLM SkillHub A real AI-agent project for your portfolio An open-source skill other developers and AI-agent users can actually use Goodies go only to the top 20 skill publishers as ranked by the judges, and mentorship only to the best skills; not every participant receives them. Who Can Participate College students across India who are interested in AI, software development, automation, research, productivity, or building useful technology. You don't have to be an AI/ML expert. What matters most is finding a useful problem and building a skill that solves it well.