Soul-driven AI agent with permission-hardened tools, token budgets, and multi-channel access. Runs 24/7 from CLI, Telegram or More.
<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0">
<channel>
<title>Mercury Blog</title>
<link>https://mercuryagent.sh/blog</link>
<description>News, engineering notes and field reports from the team building Mercury.</description>
<lastBuildDate>Wed, 11 Nov 2026 00:00:00 GMT</lastBuildDate>
<docs>https://validator.w3.org/feed/docs/rss2.html</docs>
<generator>https://github.com/jpmonette/feed</generator>
<language>en</language>
<copyright>© 2026 Cosmic Stack</copyright>
<item>
<title><![CDATA[What Is a Personal AI Assistant? 10 Real Tasks in 2026]]></title>
<link>https://mercuryagent.sh/blog/what-is-a-personal-ai-assistant</link>
<guid>https://mercuryagent.sh/blog/what-is-a-personal-ai-assistant</guid>
<pubDate>Wed, 11 Nov 2026 00:00:00 GMT</pubDate>
<description><![CDATA[An AI personal assistant acts for one person across their accounts, files, apps and connected services. Here is what it is, ten tasks it can handle, and the permissions each one needs.]]></description>
<category>Product</category>
<category>Security</category>
<category>AI Agents</category>
<category>Self-Hosted AI</category>
</item>
<item>
<title><![CDATA[9 Safest, Most Secure, and Permission-Aware AI Agents for Real-World Tasks]]></title>
<link>https://mercuryagent.sh/blog/six-ai-agents-compared-for-real-work</link>
<guid>https://mercuryagent.sh/blog/six-ai-agents-compared-for-real-work</guid>
<pubDate>Thu, 05 Nov 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Grok, ChatGPT Work, Mercury, OpenClaw, Manus, Gemini agent for work, Devin, Claude Cowork and Comet compared on memory, where they run, what they can touch, scheduling, permissions, model choice and price.]]></description>
<category>Engineering</category>
<category>Product</category>
<category>AI Agents</category>
<category>AI Agent Comparison</category>
<category>Self-Hosted AI</category>
<category>AI Agent Memory</category>
</item>
<item>
<title><![CDATA[Mercury 1.5: Mercury Desktop]]></title>
<link>https://mercuryagent.sh/blog/mercury-1-5</link>
<guid>https://mercuryagent.sh/blog/mercury-1-5</guid>
<pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Mercury 1.5.0 ships Mercury Desktop for macOS, Windows and Linux, remote access the owner approves, and updates that roll back on their own.]]></description>
<category>News</category>
<category>Releases</category>
</item>
<item>
<title><![CDATA[A tour of Mercury Desktop: one app, every Mercury you run]]></title>
<link>https://mercuryagent.sh/blog/mercury-desktop-tour</link>
<guid>https://mercuryagent.sh/blog/mercury-desktop-tour</guid>
<pubDate>Tue, 03 Nov 2026 00:00:00 GMT</pubDate>
<description><![CDATA[What's inside Mercury Desktop 1.5.0: chat, research, code, bots, memory and voice, plus connecting to several Mercury servers and keeping them all updated.]]></description>
<category>Product</category>
<category>Releases</category>
</item>
<item>
<title><![CDATA[Your agent returned 200 OK. It may still have failed]]></title>
<link>https://mercuryagent.sh/blog/agent-observability-first-wrong-step</link>
<guid>https://mercuryagent.sh/blog/agent-observability-first-wrong-step</guid>
<pubDate>Sat, 31 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[What to record in an AI agent run to find its first wrong step, and what Mercury records today. Based on OpenTelemetry's GenAI agent spans and a code review.]]></description>
<category>Engineering</category>
<category>Runtime</category>
</item>
<item>
<title><![CDATA[Running an agent overnight: what breaks first]]></title>
<link>https://mercuryagent.sh/blog/running-an-agent-overnight</link>
<guid>https://mercuryagent.sh/blog/running-an-agent-overnight</guid>
<pubDate>Tue, 27 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[We read Mercury's scheduler and budget code and reproduced one behaviour, to find what fails during long unattended runs: skipped tasks, quiet errors, overlapping jobs and budget stops.]]></description>
<category>Runtime</category>
<category>Engineering</category>
</item>
<item>
<title><![CDATA[How Optimus works: an agent that uses your computer]]></title>
<link>https://mercuryagent.sh/blog/how-optimus-works</link>
<guid>https://mercuryagent.sh/blog/how-optimus-works</guid>
<pubDate>Sat, 24 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Inside Optimus, the Mercury plugin that sees your screen, moves the mouse and types: its processes, how it looks and acts, voice control, and the safety rails.]]></description>
<category>Engineering</category>
<category>Product</category>
</item>
<item>
<title><![CDATA[Mercury 1.4: plugins, WhatsApp calls, and an agent at your desk]]></title>
<link>https://mercuryagent.sh/blog/mercury-1-4</link>
<guid>https://mercuryagent.sh/blog/mercury-1-4</guid>
<pubDate>Fri, 23 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Mercury 1.4 adds isolated plugins and two official ones: WhatsApp chat with voice calls both ways, and Optimus, which sees your screen and uses your mouse.]]></description>
<category>News</category>
<category>Releases</category>
<category>Product</category>
</item>
<item>
<title><![CDATA[The new bottleneck in AI coding isn't code generation. It's verification.]]></title>
<link>https://mercuryagent.sh/blog/ai-coding-verification-bottleneck</link>
<guid>https://mercuryagent.sh/blog/ai-coding-verification-bottleneck</guid>
<pubDate>Tue, 20 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[A Harvard working paper finds coding agents raised code output while review time rose 49%. Why verification, not generation, should decide whether an agent is useful.]]></description>
<category>Engineering</category>
<category>News</category>
</item>
<item>
<title><![CDATA[Agents exceeded their permissions: how to scope them]]></title>
<link>https://mercuryagent.sh/blog/agent-permissions-least-privilege</link>
<guid>https://mercuryagent.sh/blog/agent-permissions-least-privilege</guid>
<pubDate>Sat, 17 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[CERT-In says AI agents read live business records at Indian firms. Here is how least privilege, scoped access, approvals and audit logs should limit them.]]></description>
<category>Security</category>
<category>News</category>
</item>
<item>
<title><![CDATA[Mercury Bots: an agent fleet that runs on your machine]]></title>
<link>https://mercuryagent.sh/blog/mercury-bots</link>
<guid>https://mercuryagent.sh/blog/mercury-bots</guid>
<pubDate>Fri, 16 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Always-on agents shipped everywhere this autumn. How Mercury Bots work, what we measured, how they compare with Grok Bot, OpenAI dots, Meta Muse, OpenClaw and Hermes Agent, and the commits behind it.]]></description>
<category>Product</category>
<category>Engineering</category>
<category>Runtime</category>
</item>
<item>
<title><![CDATA[The setup wizard that quit at the first menu]]></title>
<link>https://mercuryagent.sh/blog/the-wizard-that-quit-at-the-menu</link>
<guid>https://mercuryagent.sh/blog/the-wizard-that-quit-at-the-menu</guid>
<pubDate>Thu, 15 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[For six releases, a brand-new Mercury install asked for your name, showed one menu, and exited without an error. A walkthrough of a one-tick race between two readers of the same terminal, and the test that now keeps it from coming back.]]></description>
<category>Engineering</category>
<category>Postmortem</category>
</item>
<item>
<title><![CDATA[We were running Mercury twice and didn't know it]]></title>
<link>https://mercuryagent.sh/blog/we-were-running-mercury-twice</link>
<guid>https://mercuryagent.sh/blog/we-were-running-mercury-twice</guid>
<pubDate>Wed, 14 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[Typing felt sluggish on every machine, and the terminal renderer took the blame. The real cause was a second full copy of the agent, started quietly by the launcher. How we found it, and why the fix was an architecture change rather than a performance tweak.]]></description>
<category>Engineering</category>
<category>Postmortem</category>
<category>Runtime</category>
</item>
<item>
<title><![CDATA[Mercury 1.3: an agent you can actually leave running]]></title>
<link>https://mercuryagent.sh/blog/mercury-1-3</link>
<guid>https://mercuryagent.sh/blog/mercury-1-3</guid>
<pubDate>Mon, 12 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[One background runtime that every terminal attaches to, bots that deliver real files and stay online, and a fleet you can run from Telegram. What changed in the 1.3 line and why.]]></description>
<category>News</category>
<category>Releases</category>
<category>Runtime</category>
</item>
<item>
<title><![CDATA[Asks before it acts: how Mercury's permission model works]]></title>
<link>https://mercuryagent.sh/blog/asks-before-it-acts</link>
<guid>https://mercuryagent.sh/blog/asks-before-it-acts</guid>
<pubDate>Fri, 09 Oct 2026 00:00:00 GMT</pubDate>
<description><![CDATA[An agent with a shell, a file system and network access is only useful if you can trust what it will do with them. A tour of the layers between the model and your machine: scopes, approvals, a blocklist nothing can override, an SSRF guard, and tiers for unattended bots.]]></description>
<category>Product</category>
<category>Security</category>
</item>
</channel>
</rss>