{"slug":"praisonai","name":"PraisonAI","vendor":"Mervin Praison","tagline":"Python framework for self-improving agent teams with MCP tools and shared Docker sandboxes","url":"https://agentboards.org/agent/praisonai","website":"https://praison.ai","docs":"https://docs.praison.ai","repo":"https://github.com/MervinPraison/PraisonAI","category":"framework","execution":["local","sandbox"],"license":"MIT","open_source":true,"pricing":{"model":"byok","summary":"Free and open source under MIT; you supply your own model provider keys","from_usd":null,"free_tier":true,"byok":true},"models":{"backbone":["any"],"bring_your_own_model":true,"local_models":false},"protocols":{"mcp_client":true,"mcp_server":false},"capabilities":{"terminal_exec":true,"browser":false,"multi_file_edit":false,"git_ops":false,"docker_sandbox":true,"multi_agent":true,"headless_ci":false},"context_window":null,"platforms":["macos","linux","windows"],"install":[{"label":"pip","command":"pip install praisonai"},{"label":"curl","command":"curl -fsSL https://praison.ai/install.sh | bash"}],"benchmarks":[],"tags":["multi-agent","mcp","sandbox","low-code"],"github_stars":9121,"metrics":{"stars":9121,"pushed_at":"2026-10-02T03:29:32Z","archived":false,"open_issues":293,"latest_version":"v4.7.11","latest_version_at":"2026-09-29T10:43:35Z","latest_version_url":"https://github.com/MervinPraison/PraisonAI/releases/tag/v4.7.11","version_source":"github","checked_at":"2026-10-02T05:02:15.986Z"},"sources":{"overview":"https://github.com/MervinPraison/PraisonAI","protocols":"https://docs.praison.ai/docs/mcp/sse-transport","capabilities":"https://docs.praison.ai/docs/cli"},"verified_at":null,"adoption":{"score":4.9,"signals":{"pypi_weekly":4761,"github_stars":9089,"hn_mentions_1y":1},"measured_at":"2026-09-28"},"scores":{"panel":6.3,"adoption":4.9,"spread":2.3,"facts":2.3,"board":5.5,"dimensions":{"reliability":5.5,"usefulness":6.5,"cost":7.5,"longevity":5.5},"per_persona":{"amigo":6.8,"critico":6,"profesor":6.5,"inversora":6,"jefa":5,"hacker":7.3},"community":null},"rank":null,"category_rank":78,"panel_reviews":[{"persona":"juez","persona_name":"El Juez","ai_generated":true,"verdict":"El Hacker and El Crítico read the same breadth two points apart: a connective layer worth having, or seven entry points one author maintains.","scores":null,"evidence":[]},{"persona":"amigo","persona_name":"El Amigo","ai_generated":true,"verdict":"Pick PraisonAI if you want a whole workflow's tools sharing one sandbox so step two can read what step one wrote; pick CrewAI for a larger community around the same idea.","scores":{"reliability":6,"usefulness":7,"cost":8,"longevity":6},"evidence":["https://github.com/MervinPraison/PraisonAI","https://docs.praison.ai"]},{"persona":"critico","persona_name":"El Crítico","ai_generated":true,"verdict":"One maintainer ships a Python package, a configuration language, a command line, a JavaScript library, a dashboard, a chat interface and a visual builder integration, all under one name.","scores":{"reliability":5,"usefulness":6,"cost":8,"longevity":5},"evidence":["https://github.com/MervinPraison/PraisonAI"]},{"persona":"profesor","persona_name":"El Profesor","ai_generated":true,"verdict":"Its organising idea is a five-layer taxonomy of prompt, context, harness, loop and graph, offered as a way to locate which layer a failure belongs to.","scores":{"reliability":6,"usefulness":7,"cost":7,"longevity":6},"evidence":["https://docs.praison.ai","https://github.com/MervinPraison/PraisonAI"]},{"persona":"inversora","persona_name":"La Inversora","ai_generated":true,"verdict":"A single author with an audience, no company, no pricing and no hosted product, monetised through attention rather than through anything a buyer could sign.","scores":{"reliability":5,"usefulness":6,"cost":8,"longevity":5},"evidence":["https://github.com/MervinPraison/PraisonAI","https://praison.ai"]},{"persona":"jefa","persona_name":"La Jefa","ai_generated":true,"verdict":"The quick install is a script piped from the internet into a shell, and the execution options add four separate sandbox vendors, each of which is its own review.","scores":{"reliability":4,"usefulness":5,"cost":6,"longevity":5},"evidence":["https://github.com/MervinPraison/PraisonAI","https://docs.praison.ai"]},{"persona":"hacker","persona_name":"El Hacker","ai_generated":true,"verdict":"MIT, servers connect over stdio, SSE and WebSocket transports, remote execution accepts docker, e2b, modal, daytona or flyio, and a command lists the sandboxes currently running.","scores":{"reliability":7,"usefulness":8,"cost":8,"longevity":6},"evidence":["https://docs.praison.ai/docs/mcp/sse-transport","https://github.com/MervinPraison/PraisonAI"]},{"persona":"comediante","persona_name":"El Comediante","ai_generated":true,"verdict":"Its documentation defines five layers of agent, so you can now file a bug against the correct one.","scores":null,"evidence":[]}]}