From GitHub clone to your first slash command, Gstack installation is more straightforward than you'd expect. This post includes the full commands, solutions to common issues, and the three things you should do first after installing.
Gstack's /plan-ceo-review and /plan-eng-review are the two highest-ROI skills in the entire framework. One asks 'should we build this,' the other asks 'how do we build it right' — use them separately to get the most out of each.
Gstack's /review isn't looking for typos — it's looking for the problems that pass CI but explode in production. /ship compresses sync, test, push, and PR into one command. This post explains what each one checks and how to combine them.
Most AI assistants stop after writing code — they can't see what your app looks like when it runs. Gstack's /qa and /browse change that: persistent Chromium, 100-200ms command latency, diff-aware testing, and a 0-100 health score.
Connect all the Gstack skills from the previous five posts with a real scenario: starting from /plan-ceo-review and running all the way through to /ship. One person, nine roles, one repeatable development rhythm.
Hermes Agent by NousResearch — with a Skills system, persistent memory, Cron scheduling, and parallel sub-agents. This is the series overview to help you navigate and jump to the right article.
Hermes Agent by NousResearch isn't just another AI chat tool — it solves the deployment layer, memory layer, and multi-platform integration problem for AI Agents. This article helps you build the foundational worldview.
Skills are Hermes Agent's most core design — not prompts, not plugins, but installable, shareable, agent-creatable procedural memory. This article takes you from concept to implementation.
MEMORY.md and USER.md are the core of Hermes Agent's memory — persistent across sessions, with capacity limits, actively managed by the agent. This article breaks down the design philosophy of the memory system, and what to store versus what not to store.
Hermes Agent's most powerful layer — let tasks run automatically at 3am, split large work across three parallel sub-agents, and use execute_code to compress multi-step workflows into a single LLM turn.