{"slug": "i-built-genesis-the-first-brain-for-ai-agents-agent-skills", "title": "I Built Genesis: The First Brain for AI Agents (Agent-skills)", "summary": "Developer Raiyan Islam built Genesis, a project-local cognitive layer for AI agents that preserves project knowledge across sessions. The tool, available on GitHub and Skills.sh, uses plain files to store architectural decisions, patterns, and conventions, reducing agent setup time by roughly 80% in his workflow. Genesis aims to make project knowledge version-controlled and shareable, enabling agents to start with prior context rather than rediscovering it.", "body_md": "AI coding agents are getting better at writing code.\n\nBut there is still a problem I keep seeing:\n\n**Every new agent session starts by rediscovering the project.**\n\nIt has to figure out the architecture.\n\nIt has to discover conventions.\n\nIt has to understand previous decisions.\n\nIt has to learn how the team works.\n\nAnd after the session ends, a lot of that understanding disappears.\n\nSo I built something to address that problem.\n\n**Genesis.**\n\n[GitHub — raiyanu/agent-genesis](https://github.com/raiyanu/agent-genesis)\n\n[Skill.sh — raiyanu/agent-genesis](https://www.skills.sh/raiyanu/agent-genesis/agent-genesis)\n\nGenesis is a **project-local cognitive layer for AI agents**.\n\nI think of it as the **first brain of an agent**.\n\nBefore an agent starts working on a project, Genesis gives it a structured way to:\n\nThe core idea is simple:\n\nUnderstand before acting. Learn while working. Remember for the next agent.\n\nThe workflow looks like this:\n\n```\n        Genesis\n           ↓\n      Understand\n           ↓\n         Plan\n           ↓\n        Execute\n           ↓\n          Learn\n           ↓\n        Remember\n           ↺\n```\n\nThe important part is the last step.\n\nThe agent doesn't just finish a task.\n\nIt can leave behind **knowledge that makes the next session better**.\n\nI didn't want Genesis to depend on a particular AI provider, IDE, or agent runtime.\n\nThe knowledge belongs to the **project**.\n\nThat means it can live alongside the codebase and be version controlled with Git.\n\nA simplified structure looks like:\n\n```\n.agents/\n├── SKILL.md\n└── agent_md/\n    ├── project-learn.md\n    ├── user-learn.md\n    ├── domains/\n    ├── patterns/\n    ├── decisions/\n    └── debugging/\n```\n\nThis makes project knowledge:\n\nImagine three developers working on the same project.\n\nDeveloper A teaches an AI agent about an important architectural decision.\n\nDeveloper B starts a new session tomorrow.\n\nWithout persistent knowledge, the agent may have to rediscover that decision.\n\nWith Genesis, the project can retain it.\n\n```\nDeveloper A\n     ↓\n   Agent\n     ↓\n  Discovers\n  Knowledge\n     ↓\n   Genesis\n     ↓\nProject Memory\n     ↓\nDeveloper B\n     ↓\n   Agent\n     ↓\nStarts with\nthat knowledge\n```\n\nThe project gradually becomes better at communicating with its agents.\n\nThis is an important distinction.\n\nI don't see Genesis as simply a memory folder.\n\nMemory is only one part of it.\n\nGenesis is about the **agent's relationship with the project**:\n\n**Before the task**\n\nUnderstand what already exists.\n\n**During the task**\n\nFollow the project's architecture and conventions.\n\n**After the task**\n\nIdentify what was learned and preserve knowledge that can help future work.\n\nThat creates a continuous learning loop.\n\nI intentionally kept the foundation simple.\n\nNo database.\n\nNo vector store.\n\nNo proprietary format.\n\nJust files that humans and AI agents can both understand.\n\nThat gives the knowledge the same benefits as source code:\n\n```\nGit\n ↓\nVersioning\n ↓\nReview\n ↓\nHistory\n ↓\nCollaboration\n```\n\nProject knowledge becomes part of the project itself.\n\nI think we're moving toward a world where projects won't just contain source code.\n\nThey will also contain **context for the agents working on that source code**.\n\nToday we have:\n\n```\nSource Code\nDocumentation\nTests\nConfiguration\n```\n\nI think we'll increasingly have:\n\n```\nSource Code\nDocumentation\nTests\nConfiguration\nAgent Knowledge\nAgent Instructions\nAgent Decisions\n```\n\nThe goal isn't to make agents magically smarter.\n\nIt's to make sure they **don't have to forget everything every time they start**.\n\nAfter using Genesis in my own agent workflow, I noticed a significant reduction in the time spent on project discovery and repeated context-building.\n\nFor my workflow, it was **roughly 80% faster** to get an agent into a productive state.\n\nThe important part isn't that Genesis makes the AI model itself faster.\n\nIt reduces the repeated work around the model:\n\n```\nWithout Genesis\nAgent → Explore → Rediscover(grepping 10+ function and reading 1200+ lines of code) → Understand → Start Working\n\nWith Genesis\nAgent → Load Context → Understand → Start Working\n```\n\nThat difference becomes especially noticeable when working across multiple sessions or switching between CLI and IDE agents.\n\nThe agent doesn't have to start from zero every time.\n\nIt starts with what the project has already learned.\n\nFor every Module, First prompt was slower but second prompt doesnt matter if it's different session or account or agent or system it become faster than before. for my ecommerce project it improved agent time in searching for code from 2-3 minute to 20 seconds.\n\nThe name comes from the idea of a beginning.\n\nGenesis is the point where the agent starts.\n\nNot with assumptions.\n\nNot with a blank context.\n\nBut with the knowledge the project has already accumulated.\n\nEvery agent session should start smarter than the previous one.\n\nGenesis is my first step toward that idea.\n\nThe project is open source, and I'm interested in seeing how this concept evolves with the rapidly changing AI-agent ecosystem.\n\nIf you're building CLI agents, IDE agents, coding agents, or autonomous development workflows, I'd love to hear how you're solving the **persistent context problem**.", "url": "https://wpnews.pro/news/i-built-genesis-the-first-brain-for-ai-agents-agent-skills", "canonical_source": "https://dev.to/raiyanu/i-built-genesis-the-first-brain-for-ai-agents-agent-skills-41f4", "published_at": "2026-08-10 13:35:36+00:00", "updated_at": "2026-08-10 13:49:27.982397+00:00", "lang": "en", "topics": ["ai-agents", "developer-tools", "artificial-intelligence"], "entities": ["Raiyan Islam", "Genesis", "GitHub"], "alternates": {"html": "https://wpnews.pro/news/i-built-genesis-the-first-brain-for-ai-agents-agent-skills", "markdown": "https://wpnews.pro/news/i-built-genesis-the-first-brain-for-ai-agents-agent-skills.md", "text": "https://wpnews.pro/news/i-built-genesis-the-first-brain-for-ai-agents-agent-skills.txt", "jsonld": "https://wpnews.pro/news/i-built-genesis-the-first-brain-for-ai-agents-agent-skills.jsonld"}}