{"slug": "building-persistent-agents", "title": "Building Persistent Agents", "summary": "A sponsored post by Omar Sar (omarsar0) describes testing Fulcra Dynamics' shared context layer to hand off work between two coding agents, with Claude Code saving its progress, decisions and next steps to a Fulcra workspace and a fresh Codex session then reading that workspace and completing the final two steps without redoing the first, with all tests passing. The post argues that persistent memory outside any single harness is needed for long-running agents that can run for days, restart, switch models or exhaust context, and notes that four Gemini 4 Argon agents beat Minecraft in 27 hours while sharing lessons through a shared memory. Fulcra's agent skills can be added with 'npx skills add fulcradynamics/agent-skills', and the post discloses that Fulcra partnered on it.", "body_md": "More and more, I write about the importance of owning your intelligence stack.\n\nThat includes owning your agent's memory. Beyond the harness and models, memory is a key part of the intelligence stack you want to manage well to unlock agents' frontier capabilities.\n\nHow does this look in practice and why does it matter?\n\nLet's focus on persistent agents, which is one of the most exciting new applications today.\n\nPersistence has become a core capability of recent models like Opus 5.5, GPT -6 Astra, and Gemini 4 Argon.\n\nRecently, four Gemini 4 Argon agents beat Minecraft in 27 hours. Along the way, they recovered items, revised strategies, and shared lessons through a shared memory.\n\nModels are getting better at choosing what to persist, and how and when to use it. What they persist needs a place that outlives any one session, and you should fully control that.\n\nRecently, I have been building and testing my own personal agents. What I have found is that memory is key to what makes them useful.\n\nSpecifically, I want personal agents for long-running tasks.\n\nIn the standard setup, memory usually lives inside one harness on one machine, so when I switch to a different agent, it can't see what the first one learned.\n\nI have been exploring a few solutions.\n\nRecently, I have been testing [@FulcraDynamics](https://x.com/FulcraDynamics), which gives your agents a shared context layer you own.\n\nOne agent writes to it, and the next one picks up from there, across tools and sessions.\n\nSo I tested it on a real handoff between two coding agents.\n\nI gave Claude Code a small coding task and had it save its progress, decisions, and next steps to a shared Fulcra workspace after the first step. Then I opened a fresh Codex session and only told it the workspace name.\n\nCodex read the workspace, figured out what was done and what was left, and finished the last two steps without redoing the first. All the tests passed, and I never pasted anything between the two agents.\n\nThis is extremely useful for long-running agents.\n\nLong-horizon agents can run for days, get restarted, switch models, and run out of context. When their state lives in Fulcra, a fresh session reads where things stand and picks up the work from there. This lets work persist, which you need to build agents for complex, long-running tasks.\n\nYou can also decide what each agent sees. You can share one folder with a person or their agent and keep everything else private.\n\nIt's your context, and you can export or delete it at any time.\n\nYou can start by adding their agent skills with npx skills add fulcradynamics/agent-skills\n\nTry Fulcra here: [https://www.fulcradynamics.com/product/start-for-free?utm_source=x&utm_medium=paid_social&utm_campaign=omarsar0](https://www.fulcradynamics.com/product/start-for-free?utm_source=x&utm_medium=paid_social&utm_campaign=omarsar0)\n\nThanks to Fulcra for partnering on this post.", "url": "https://wpnews.pro/news/building-persistent-agents", "canonical_source": "https://twitter.com/omarsar0/status/2112573314631708736", "published_at": "2026-10-20 16:02:11+00:00", "updated_at": "2026-10-20 16:17:09.107733+00:00", "lang": "en", "topics": ["ai-agents", "ai-tools", "large-language-models"], "entities": ["Fulcra Dynamics", "Claude Code", "Codex", "Gemini 4 Argon", "Omar Sar", "Opus 5.5", "GPT-6 Astra", "Minecraft"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/building-persistent-agents", "markdown": "https://wpnews.pro/news/building-persistent-agents.md", "text": "https://wpnews.pro/news/building-persistent-agents.txt", "jsonld": "https://wpnews.pro/news/building-persistent-agents.jsonld"}}