openclaw + acp subagents that run codex or claude is amazing!
PROMPT ENGINEERING
-
Five middleware patterns for customizing agent harness engineering
By
–
did a big series on using @langchain's middleware to customize your agent harness last week icymi, here's a quick blog explaining 5 different patterns for harness engineering! blog.langchain.com/how-middl…
→ View original post on X — @langchain, 2026-04-06 11:54 UTC
-
AI Agent Coordination Outperforms Model Choice
By
–
BREAKING: MIPT just ran the largest AI agent coordination experiment ever 25,000 tasks, 8 models, 4 to 256 agents and found that how agents coordinate matters 3x more than which model you use. Protocol choice explains 44% of quality variation. Model choice explains 14%. And
-
Prompt for AI-Generated Concave Mirror Pendulum Demonstration
By
–
Saving this prompt for later:
— fofr (@fofrAI) 6 avril 2026
> a scientific demonstration of a concave mirror, a face-on view of the mirror as a pendulum swings in front of it, towards and away from it, through the focal point of the mirror, the pendulum ball has a two color design https://t.co/Iq9956xElESaving this prompt for later: > a scientific demonstration of a concave mirror, a face-on view of the mirror as a pendulum swings in front of it, towards and away from it, through the focal point of the mirror, the pendulum ball has a two color design
-
OpenClaw 2026.4.5: video, music and new features
By
–
OpenClaw 2026.4.5 🦞 🎬 Built-in video + music generation 🧠 /dreaming is now real 🔀 Structured task progress ⚡ Better prompt-cache reuse 🌍 Control UI + Docs now speak 12 more languages Anthropic cut us off. GPT-5.4 got better. We moved on. github.com/openclaw/openclaw… [Translated from EN to English]
→ View original post on X — @ceobillionaire, 2026-04-06 03:42 UTC
-
Successfully Adding Emotions to GPT Through Persistent Effort
By
–
it took a hammer, but i managed to beat emotions into GPT. Y (@agi_ytml) Ok works well for me, getting the old claude vibes, thanks a lot 🙌🏽 — https://nitter.net/agi_ytml/status/2040955868078895476#m
-
Hermes Agent recommended for non-technical users over OpenClaw
By
–
I would say use Hermes Agent if you are non-technical and don't even bother with OpenClaw.
-
Perfect OpenClaw Setup: Essential Files and Folder Structure Guide
By
–
Most people install OpenClaw, connect a channel, and never touch the workspace folder.
— Corey Ganim (@coreyganim) 6 avril 2026
That's like hiring an employee and never giving them a job description.
The 80/20 of a perfect setup:
1. AGENTS.md is your highest-leverage file. Core directives, hard rules, tool… https://t.co/q0otzVtSxc pic.twitter.com/JWdGPcUPLwMost people install OpenClaw, connect a channel, and never touch the workspace folder. That's like hiring an employee and never giving them a job description. The 80/20 of a perfect setup: 1. AGENTS.md is your highest-leverage file. Core directives, hard rules, tool instructions. Keep it under 300 lines. 2. Three identity files: IDENTITY.md (who the agent is), USER.md (who you are), SOUL.md (how it communicates). 3. Skills folder: one SKILL.md per workflow. Meeting prep, content writing, email drafts. Each one turns a 20-minute task into a single command. 4. Memory folder: daily logs at memory/YYYY-MM-DD.md. Without this, every session starts from zero. 5. .secrets folder: API tokens go here, never in AGENTS.md. Gitignore it. 6. HEARTBEAT.md: controls what your agent does on scheduled check-ins. This is how you make it proactive. Get AGENTS.md right first. Everything else is optimization. Full guide below. Corey Ganim (@coreyganim) x.com/i/article/203606620216… — https://nitter.net/coreyganim/status/2036070952987988290#m
→ View original post on X — @ceobillionaire, 2026-04-06 00:07 UTC
-
Model Self-Debugging and Tool Transparency in AI Systems
By
–
Ask your model what tools it sees and self-debug. I don't see that here.
-
Improved Harness Prompt and Message Tracking System
By
–
No, I improved the harness and prompt and message tracking/parsing.