Techimus Prime

@HuggingModels github.com/cosmicstack-la…

When people finally realize Hermes and OpenClaw are just toys compared to Mercury Agent.

@m_shehzen @ryanvogel +1

@ryanvogel Yet, open code prevents itself forking and adding new features.

OpenCode is completely open source now! github.com/anomalyco/open…

@noobaibuilder @GithubProjects

@TechimusPrime @GithubProjects @TechimusPrime is this real lmao?

Paste your GitHub streak chart screenshot.

Finally pre-ordered GTAVI. After almost 14 yrs. Thank you @RockstarGames

ask a hacker to go through your vibe coded app

@mercury__agent Plus the amount of data context send to each LLM call is never bloated as dumb wall

The AI industry is optimizing for larger context windows. Mercury is optimizing for needing fewer of them. Better memory beats bigger prompts. ☿ Mercury Agent

We spent years optimizing databases for storage. Now we’re optimizing agents for memory. The challenge isn’t storing information. The challenge is retrieving the right information at the right time with the smallest possible context footprint. That’s where most agent architectures break. ☿ Mercury Agent

@GithubProjects GLM 5.1 or 5.2

What model do you use for coding?

@Ctrl_Alt_Zaid @m_shehzen Lets go

GLM 5.2 is so good that we are considering it as new default provider for Mercury Agent

Mercury Agent v1.1.13 is here. We call it: Chatty Mercury. Mercury now works across 6 channels: • CLI • Web Dashboard • Telegram • Discord • Slack • Signal The goal wasn’t to add more integrations. The goal was to let you talk to your agent wherever you already work. This release also fixes one of the most frustrating agent problems: silent failures. Tasks no longer disappear without explanation. Loop detection, stall detection, crash recovery, channel error reporting, and watchdog recovery now ensure Mercury tells you exactly what happened instead of quietly stopping. We also cleaned up the CLI experience: • Heartbeats update in place • Step logs stay compact • Better processing indicators • Improved daemon shutdown • Fixed Ollama local compatibility Agents should be reliable. If they fail, they should tell you why. If they crash, they should tell you what happened. If they run long, they should show progress. v1.1.13 moves Mercury closer to that standard. Your agent. Your rules.

Mercury Agent v1.1.13 is here. Most AI agents are powerful. Few actually remember. Mercury Agent’s Second Brain gives agents persistent relational memory, allowing them to remember people, projects, decisions, and context across sessions without repeatedly consuming tokens to rebuild state. Built for long-term workflows, not one-off conversations. Highlights: • Second Brain relational memory • Scoped Permissions with approval-first execution • Unified Skill Management • Token Saver for smarter context usage • 31 built-in tools • Long-running autonomous workflows • Millisecond memory retrieval • Bring your own OpenAI-compatible models and endpoints Faster recall. Lower token usage. More context retained. Because the future isn’t bigger context windows. It’s agents that remember. Your agent. Your rules.

Mercury Agent v1.1.13 is here. We call it: Chatty Mercury. Mercury now works across 6 channels: • CLI • Web Dashboard • Telegram • Discord • Slack • Signal The goal wasn’t to add more integrations. The goal was to let you talk to your agent wherever you already work. This release also fixes one of the most frustrating agent problems: silent failures. Tasks no longer disappear without explanation. Loop detection, stall detection, crash recovery, channel error reporting, and watchdog recovery now ensure Mercury tells you exactly what happened instead of quietly stopping. We also cleaned up the CLI experience: • Heartbeats update in place • Step logs stay compact • Better processing indicators • Improved daemon shutdown • Fixed Ollama local compatibility Agents should be reliable. If they fail, they should tell you why. If they crash, they should tell you what happened. If they run long, they should show progress. v1.1.13 moves Mercury closer to that standard. Your agent. Your rules.

Mercury now speaks Discord, Slack & Signal. ☿ v1.1.13 Chatty is live. Three new gateways. One agent. More features and improvements Details below 🧵

Mercury Agent v1.1.13 is here. We call it: Chatty Mercury. Mercury now works across 6 channels: • CLI • Web Dashboard • Telegram • Discord • Slack • Signal The goal wasn’t to add more integrations. The goal was to let you talk to your agent wherever you already work. This release also fixes one of the most frustrating agent problems: silent failures. Tasks no longer disappear without explanation. Loop detection, stall detection, crash recovery, channel error reporting, and watchdog recovery now ensure Mercury tells you exactly what happened instead of quietly stopping. We also cleaned up the CLI experience: • Heartbeats update in place • Step logs stay compact • Better processing indicators • Improved daemon shutdown • Fixed Ollama local compatibility Agents should be reliable. If they fail, they should tell you why. If they crash, they should tell you what happened. If they run long, they should show progress. v1.1.13 moves Mercury closer to that standard. Your agent. Your rules.

@mercury__agent 🦾❤️

☢️ Nobody talks about this: You only need loops when your agent is bad at giving results at first try.

@ClaudioDrews25 Dropping v1.1.13 with more capabilities and provider and gateway support

@TechimusPrime Mercury v1.1.12 uses OpenAI's Responses API (/v1/responses) exclusively? 🤔

Next update of mercury agent has got you at lightning speed. > eliminated all 12 silent task failure paths

Nice idea till we get endless loop without an exit condition.