No-Code Agentist

A daily digest of practical AI agent workflows for non-developers. We isolate actionable no-code guides from viral hype. Scored against human-defined standards.

Latest Oct 2 Oct 1 Sep 30 Sep 29 Sep 28 Sep 27 Sep 26 Sep 25 Sep 24 Sep 23 Sep 22 Sep 21 Sep 20 Sep 19 Sep 18 Sep 17 Sep 16 Sep 15 Sep 14 Sep 13 Sep 12 Sep 11 Sep 10 Sep 9 Sep 8 Sep 7 Sep 6 Sep 5 Sep 4 Sep 3 Sep 2 Sep 1 Aug 31 Aug 30 Aug 29 Aug 28 Aug 27 Aug 26 Aug 25 Aug 24 Aug 23 Aug 22 Aug 21 Aug 20 Aug 19 Aug 18 Aug 17 Aug 16 Aug 15 Aug 14 Aug 13 Aug 12 Aug 11 Aug 10 Aug 9 Aug 8 Aug 7 Aug 6 Aug 5 Aug 4 Aug 3 Aug 2 Aug 1 Jul 31 Jul 30 Jul 29 Jul 28 Jul 27 Jul 26 Jul 25 Jul 24 Jul 23 Jul 22 Jul 21 Jul 20 Jul 19 Jul 18 Jul 17 Jul 16 Jul 15 Jul 14 Jul 13 Jul 12 Jul 11 Jul 10 Jul 9 Jul 8 Jul 7 Jul 6 Jul 5 Jul 4 Jul 3 Jul 2 Jul 1 Jun 30 Jun 29 Jun 28 Jun 27 Jun 26 Jun 25 Jun 24 Jun 23 Jun 22 Jun 21 Jun 20 Jun 19 Jun 18 Jun 17 Jun 16 Jun 15 Jun 14 Jun 13 Jun 12 Jun 11 Jun 10 Jun 9 Jun 8 Jun 7 Jun 6 Jun 5 Jun 4 Jun 3 Jun 2 Jun 1 May 31 May 30 May 29 May 28 May 27 May 26 May 25 May 24 May 23 May 22 May 21 May 20 May 19 May 18 May 17 May 16 May 15 May 14 May 13 May 12 May 11 May 10 May 9 May 8 May 7 May 6 May 5 May 4 May 3 May 2 May 1 Apr 30 Apr 29 Apr 28 Apr 27 Apr 26 Apr 25 Apr 24 Apr 23 Apr 22 Apr 21 Apr 20 Apr 19 Apr 18 Apr 17 Apr 16 Apr 15 Apr 14 Apr 13 Apr 12 Apr 11 Apr 10 Apr 9 Apr 8 Apr 7 Apr 6 Apr 5 Apr 4 Apr 3 Apr 2 Apr 1 Mar 31 Mar 30 Mar 29 Mar 28 Mar 27 Mar 26 Mar 25 Mar 24 Mar 23 Mar 22 Mar 21 Mar 20 Mar 18 Mar 17 Mar 16 Mar 15 Mar 14 Mar 13 Mar 12 Mar 11 Mar 10 Mar 9 Mar 8 Mar 7 Mar 6 Mar 5 Mar 4 Mar 3 Mar 2 Mar 1 Feb 28 Feb 27 Feb 26

Daily Summary

4 curated | 4 evaluated

The no-code agent landscape shifted toward composable primitives and persistent memory systems, with that self-organize task decomposition without hardcoded workflows, as simple markdown files with YAML frontmatter, and community builders demonstrating that give agents workflow history and decision context across multiple AI platforms.

•jarvana_sv released Hermes v0.17 with Kanban swarms that spawn a root agent plus parallel workers and verifiers on a shared board, featuring persistent profiles and background curators that autonomously prune memory and maintain state.

•_vmlops highlighted Anthropic's open-sourced skills system where agents load markdown files with YAML frontmatter dynamically at runtime, eliminating the need for fine-tuning or retraining while powering Claude's document creation features.

•tonysimons_ tested MOA 2.0 for Hermes Agent, which uses a panel of reference models followed by an aggregator that synthesizes responses inside the normal agent loop, achieving 8% higher scores than Opus alone.

Jun 28, 7:42 AM

https://t.co/rfAmlPVJ3o Hermes v0.17 Turns One Agent Into Full Kanban Team New update drops Kanban swarms, persistent profiles, and background curators. One prompt spawns root agent + parallel workers + verifier that hand off tasks on shared board. How it works: profiles system lets multiple isolated agents share one host. Background curator runs autonomously — prunes memory, consolidates skills, maintains state without manual prompting. No CrewAI boilerplate. No hardcoded workflows. Just self-organizing task decomposition. Example: feed it a PR. Root agent breaks it into subtasks, spawns workers for research + code + review, then verifier checks output. All tracked on Kanban board. Desktop reach added too — agents can interact with local apps, not just terminal. If you're running local agents or comparing open-source stacks, this update changes the trade-off between control and automation. Who's testing the new curator release? https://t.co/LkNCnTuZV1 https://t.co/rfAmlPVJ3o

agentic-orchestration100

high-signal-utility100

no-code-feasibility50

Jun 28, 8:15 AM

ANTHROPIC OPEN-SOURCED THE SKILLS SYSTEM POWERING CLAUDE this is what "agent memory" actually looks like in production skills are folders. a SKILL. md file. yaml frontmatter + instructions. that's the whole primitive → claude loads them dynamically at runtime → no fine-tuning. no retraining. just a markdown file → works in claude. ai, claude code, and the api the repo ships with skills across creative, technical, and enterprise workflows plus the exact docx, pdf, pptx, and xlsx skills running inside claude's document creation feature right now 149k stars. anthropic is making the whole system inspectable the real unlock....? you can write your own skill in 5 minutes. drop a SKILL.md into a folder, define what claude should do, point it at the api that's how you build repeatable agent behavior without touching model weights https://t.co/urT5vN46Io

Jun 27, 8:27 PM

Last night @NousResearch dropped MOA 2.0 for Hermes Agent. The default preset scores 8% higher than Opus alone. I spent today running it through real workflows so you don't have to. Short version: it's a panel of advisors, not a gamble on one brain. Reference models think first. An aggregator synthesizes. The whole thing works inside the normal agent loop. No glue code, no custom routing -- the way it should be, damnit. Grab a drink and buckle up, buttercup. Deep dive below. 🤘

Quoted tonysimons_:"https://t.co/T4jKkKsJwb"

Jun 27, 5:48 PM

THIS GUY BUILT A FREE OBSIDIAN SETUP THAT GIVES HIS AI AGENTS MEMORY Most agents are not failing because the model is weak. They fail because every task starts like a stranger walking into your house. > No notes. > No workflow history. > No tool map. > No memory of past decisions. > No idea what you were already building. In the video, he shows the opposite setup: Obsidian as the local memory layer for his agent stack. His notes, workflows, tools, memories, and ideas sit in one graph. Then Hermes, OpenClaw, Antigravity, Claude, Codex, Gemini, and Free Claude Code are shown plugged into that graph instead of relying only on whatever gets pasted into a prompt. That is the useful part. > Not the purple lighting. > Not the graph looking cool. > Not the “100 times better” caption. The mechanism is simple: 1. Put the important context in Obsidian 2. Keep it structured enough to retrieve 3. Let agents inspect the relevant notes before acting 4. Let the useful outputs and decisions become new context 5. Repeat until the vault becomes operational memory A normal agent can complete a task. An agent with your Obsidian context can see your naming conventions, active projects, preferred tools, unfinished ideas, recurring constraints, and past choices. That is a very different kind of automation. It stops feeling like “prompt a chatbot again” and starts feeling closer to delegation. Caveat: A messy vault does not become smart because an agent can read it. If the notes are vague, duplicated, outdated, or full of junk, the agent just retrieves better-looking confusion. The edge is not “AI remembers everything.” The edge is giving your agents one durable place to inspect your world before they touch it. Obsidian is not just a second brain for humans anymore. It can become the memory interface for the agent stack.

Quoted kocer_eth:"https://t.co/XTQhT6Qpjl"