{"data":{"slug":"microsoft-agent-lightning","name":"agent-lightning","tagline":"The absolute trainer to light up AI agents","github_url":"https://github.com/microsoft/agent-lightning","owner":"microsoft","repo":"agent-lightning","owner_avatar_url":"https://avatars.githubusercontent.com/u/6154722?v=4","primary_language":"Python","stars":17500,"forks":1541,"topics":["agent","agentic-ai","llm","mlops","reinforcement-learning"],"archived":false,"github_pushed_at":"2026-08-19T04:24:29+00:00","maintenance_label":"Very active","stars_delta_30d":104,"url":"https://www.graphcanon.com/tools/microsoft-agent-lightning","markdown_url":"https://www.graphcanon.com/tools/microsoft-agent-lightning.md","api_url":"https://www.graphcanon.com/api/graphcanon/tools/microsoft-agent-lightning","graph_url":"https://www.graphcanon.com/api/graphcanon/graph?tool=microsoft-agent-lightning","description":"The absolute trainer to light up AI agents.","homepage_url":"https://microsoft.github.io/agent-lightning/","license":"MIT","open_issues":156,"watchers":82,"ai_summary":"A Python-based tool designed for training and operating AI agents, with support for MLOps, reinforcement learning, and more.","readme_excerpt":"## ⚡ Installation\n\nThe following is an example installation on a CUDA 13.0 machine:\n\n```bash\ncd <this-repo>\nuv sync\nbash scripts/setup_verl.sh 0.8.0 cu130\n```\n\nSee the [Installation Guide](https://microsoft.github.io/agent-lightning/stable/1-installation/) for details.\n\n---\n\n## ⚡ License\n\nAgent Lightning v1.0 is released under the [MIT License](LICENSE).","github_created_at":"2025-06-18T07:28:45+00:00","created_at":"2026-07-07T17:38:52.369352+00:00","updated_at":"2026-08-19T06:01:07.959871+00:00","categories":[{"slug":"ai-agents","name":"AI Agents","url":"https://www.graphcanon.com/categories/ai-agents","markdown_url":"https://www.graphcanon.com/categories/ai-agents.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/ai-agents"},{"slug":"model-training","name":"Model Training","url":"https://www.graphcanon.com/categories/model-training","markdown_url":"https://www.graphcanon.com/categories/model-training.md","api_url":"https://www.graphcanon.com/api/graphcanon/categories/model-training"}],"tags":[{"slug":"agent","name":"agent"},{"slug":"agentic-ai","name":"agentic-ai"},{"slug":"llm","name":"llm"},{"slug":"mlops","name":"mlops"},{"slug":"reinforcement-learning","name":"reinforcement-learning"}],"trust":{"provenance":{"is_fork":false,"github_id":1004147641,"owner_type":"Organization","methodology":"github_public_v1","parent_repo":null,"near_duplicate_slugs":[]},"computed_at":"2026-08-19T06:01:07.162Z","maintenance":{"label":"Very active","score":96,"methodology":"github_public_v1","releases_90d":1,"days_since_push":0,"last_release_at":"2026-08-17T04:48:12Z","stars_delta_30d":104,"open_issues_delta_30d":3},"security_summary":{"status":"no_lockfile","scanner":null,"low_count":0,"high_count":0,"last_scan_at":"2026-08-09T04:00:42.244Z","medium_count":0,"scan_profile":"none","critical_count":0}},"capability_facts":{"scan":{"source":"repo_scan","observed_at":"2026-08-19T06:01:07.624Z"},"has_cli":{"value":true,"source":"pyproject.toml:[project.scripts]","observed_at":"2026-08-19T06:01:07.624Z"},"languages":{"value":["python"],"source":"github.language+pyproject.toml","observed_at":"2026-08-19T06:01:07.624Z"},"license_spdx":{"value":"MIT","source":"github.license","observed_at":"2026-08-19T06:01:07.624Z"}},"decision_facts":{"hosting":null,"pricing":null,"requirements":null,"constraints":null,"when_to_use":["When you're dealing with projects that require continuous integration and deployment workflows (CI/CD), as agent-lightning supports MLOps practices which facilitate this.","If your project involves reinforcement learning tasks or simulation environments where an easy-to-use tool for managing agents is needed."],"when_not_to_use":["Avoid using agent-lightning if you're working in a language other than Python, given that it's only available as a Python package.","It might not be the best fit for projects where stability over cutting-edge features is prioritized. Its development focus leans towards experimental functionalities, with frequent updates to the Test"],"source":"enrich:decision_facts","observed_at":"2026-07-11T14:38:46.281Z"},"constraint_facets":null,"decision_summary":[{"label":"Adopt for","value":"Detailed examination of agent-lightning reveals its strong points in model training, specifically its support for reinforcement learning and MLOps."}]}}