Deep dives into working agents.

A Formal Language for Writing Agent Instructions: Matt Pocock's Skills v1.2.0

A twelve-minute walkthrough of five new skills and a major infrastructure update for the most-starred agent skills repository on GitHub — now 204K stars, MIT-licensed, and an official Claude Code marketplace plugin alongside new Codex sidecar files. The headliner is /wait-what for Opus verbosity, /grill-me with multi-question dependency graphs, and /wizard for deterministic provisioning — but the skill nobody called exciting is the one worth reading: /writing-for-agents, an 81-line formal language for designing instructions agents consume — context pointers, the two loads, information hierarchy, completion criteria, leading words, negation as a failure mode, and pruning.

Forgetting as the Compression Step: Tencent's Agent Memory Engine, Checked

Tencent open-sourced an MIT agent-memory plugin that improves results by throwing information away: verbose tool logs fold into a Mermaid graph with node_ids you can grep back, and long-term memory climbs an L0-to-L3 pyramid modelled on Tulving's 1972 consolidation arrow. The repo is real (10,811 stars), the architecture is documented, and the benchmark table publishes raw token counts - so the presenter divides them on camera and finds one row that does not reconcile: SWE-bench prints -33.09% but computes to -31.63%. He also flags that every number is vendor-produced with no independent reproduction, and names the open GitHub issue where prompt-cache invalidation claws back part of the saving. The gains shrink as tasks get harder: +51.52% on WideSearch, +9.93% on SWE-bench. The video's own description gets this wrong; the video does not.

Wayfinder: Planning as a Map You Can Leave and Return To

Most agent planning tools die at the edge of a context window, and the real cost is not lost notes but trimmed ambition. Matt Pocock's Wayfinder skill moves the plan out of the session and into the issue tracker as a map of decision tickets, so work outlives any single conversation. Reading the skill source alongside the video surfaces two constraints the video never states: every ticket is either human-in-the-loop or agent-driven, and the skill forbids an agent standing in for the human's side of a discussion — the rule that keeps a decision map from looking complete while encoding nobody's judgement.

This Skill Just Made Kimi K3 A 10x Better Designer

AI LABS tests Kimi K3's frontend design capabilities against Claude and GPT, then introduces Hallmark — an anti-AI-slop design skill with 58 detection gates, 100+ references, and four verbs (default, audit, redesign, study) that strips every model's default patterns and produces intentional, original designs.

System Prompts Leaks GitHub Explained: What AI Agent Prompts Reveal

Alex Hitt breaks down the system_prompts_leaks repository — 43,600+ stars on GitHub — revealing how researchers verify leaked AI system prompts, the four-layer architecture inside commercial models, and why prompt obfuscation as a competitive moat is dead.

Don't Waste Time on Specs — /prototype Instead

Matt Pocock argues that spec-driven AI development wastes effort. His /prototype skill creates throwaway code artifacts at higher fidelity than discussion, letting you iterate on live UIs and logic before handing off to production — bridging the gap between spec and working software.

What Would It Cost to Run Claude Fable 5 Locally?

Kimi K3: An Open Source #1 — Coding Tests vs Fable 5, Opus 4.8 & GPT-5.6 Sol

Codex vs Fable: Which AI Agent Picked the Better Problem?

Nate B Jones gave Fable and Codex the same open brief: inspect his real business and build the automation that matters. Fable chose the higher-leverage strategic problem; Codex chose the safe, finishable one. Neither was wrong — they optimized for different things. The real unlock: letting AI help choose the problem, not just the tool.

I Replaced Hermes Agent And OpenClaw With This

mattpocock/skills: Learn the Whole Flow, End-to-End

First proper tutorial for the 162K-star skills repo (7.5M downloads, 38 skills). Complete main flow walkthrough: install via npx, configure issue tracker (GitHub/Jira/Linear/local), grill-with-docs interview, to-spec compression, to-tickets breakdown (each ticket = one context window), implement with auto code review via sub-agents. Smart zone concept (~140K tokens). Only 660 tokens context load despite 38 skills.

Claude Fable, Claude Tag, and Anthropic's Culture — Cat Wu & Thariq Shihipar ft Simon Willison

Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind

Philipp Schmid (Staff Engineer, Google DeepMind) reveals that almost none of 50,000+ indexed skills have evals. Covers the full lifecycle: capability vs preference skills, 8 writing tips, a practical eval harness (117 test cases, ~90% accuracy for Gemini Interactions API), and Google DeepMind's CI-style eval workflow where skill changes don't merge without improving tests.

This Skill Makes Claude BRUTALLY Honest About Your Ideas

Brad built a /focus-group skill that clones your real customers into AI agents. Paste any landing page, offer, or email and get honest reactions — not yes-man validation. Two-round system with persistent memory, real customer quotes, and actionable holdout analysis.

12 Hidden Settings to 10x Your Claude Code

Claude Code has hundreds of settings but you only need 12. Notification sounds, mobile push, smart allow lists (/fewer-permission-prompt), deny rules (.env + git push), Opus Plan + Sonnet Build (halve usage), effort level sweet spot (max is a trap), privacy/telemetry off, history retention 365 days, auto-compact at 75%, status line monitoring, prompt stashing (Ctrl+S), and removing AI commit attribution. Plus bonus: the prompting shift from steps to outcomes.

Skills v1.1: /wayfinder, /research, /implement, /to-spec, /to-tickets

Matt Pocock releases Skills v1.1 — major update (160K stars, 7M downloads). Star feature: Wayfinder for planning large projects via GitHub Issues with blocking relationships. Plus: complete SDLC flow (Grill→Spec→Tickets→Implement→Code Review), Martin Fowler refactoring smells in code review, and grilling improvements.

OpenCV 5 Is Here — The Biggest Update Since 2018 (We Tested It)

Better Stack tests OpenCV 5's brand-new DNN engine — ONNX operator coverage jumped from 22% to 80%, enabling YOLO, stable diffusion, and VLMs to run natively on CPU with zero extra dependencies. Four hands-on demos included.

The Best AI Agents Need Less Code Than You Think — LangSmith Engine Deep Dive

Deep dive into LangSmith Engine — an always-on agent that hunts production trace failures, clusters issues, and opens fix PRs. Org-chart architecture with 4 sub-agents (screener, verifier, issue creator, fix generator). Self-improving: Engine runs on its own traces. From noisy PRs to an issue inbox.

Building Great Agent Skills: The Missing Manual

Matt Pocock presents a four-part checklist (Trigger, Structure, Steering, Pruning) for escaping "skill hell" — a structured framework for building, evaluating, and maintaining high-quality agent skills.

GLM-5.2 vs MiniMax-M3 — Opus Has REAL COMPETITION (Model Stacking)

IndyDevDan analyzes GLM-5.2 and MiniMax-M3 — the open-weight models now competitive with Opus 4.8. Introduces a three-tier model stack framework, the trade-off triangle (performance, speed, cost), and the 5× cost curve between tiers for engineering and product agents.

Google OKF: The Simple Folder That Gives AI Agents Your Entire Company Data

Google's Open Knowledge Format (OKF) is a shockingly simple open spec — a folder of Markdown files in Git that gives AI agents your company's institutional memory. No database, no vectors, no platform — just YAML frontmatter + Markdown body, with progressive disclosure and an enrichment agent that auto-drafts files from BigQuery.

Claude Code Has Quietly Evolved (People Haven't Noticed)

Simon Scrapes breaks down the three biggest paradigm shifts in Claude Code usage: rethinking memory with community plugins like MemArch and GBrain, leveraging /effort and Ultra Code for multi-agent planning workflows, and automating repetitive tasks with /goal and /routines.

PLANS For Fable 5 — Rebuilding the /Plan Skill for Mythos Class Models

IndyDevDan rebuilds his /plan meta skill from scratch for Fable 5 and Mythos class models. A 63-minute deep devlog covering property-based engineering, HTML-first plan templates, five unified workflows, embedded GPT Image 2 generation, and plans as living artifacts — all built around the thesis that great planning is great engineering.

You NEED to Try These 7 Loops

Matthew Berman breaks down AI coding agent loops — autonomous goal-driven iteration cycles that remove humans from the development loop. Seven concrete loop patterns from sub-50ms page loads to overnight documentation sweeps, with practical prompts you can copy directly into Codex or Claude Code.