Resources & Links
Everything you need to get started with OpenClaw, set up memory, and go deeper on the research behind this talk.
🦞 OpenClaw Core
The platform itself — start here.
OpenClaw Website
The official site — overview, features, and getting started.
openclaw.ai →OpenClaw Documentation
Full reference: configuration, plugins, hooks, agents, tools, and deployment guides.
docs.openclaw.ai →OpenClaw GitHub Repository
The source code. Star it, fork it, read the issues for real-world problems and solutions.
github.com/openclaw →OpenClaw Discord
The community. Ask questions, share configs, see what others are building.
Join Discord →ClawHub — Skills Marketplace
5,700+ community skills. Browse, install, or publish your own.
clawhub.com →🧠 Memory Plugins
The plugins we discussed in the talk — the memory stack.
LCM — Lossless Claw
Context management via DAG-based summarization. Prevents information loss during compaction. By Martian Engineering.
GitHub →Graphiti — Knowledge Graph Framework
The graph engine behind TrueMem. Entity resolution, temporal awareness, hybrid search. By Zep AI.
GitHub →TrueMem — Semantic Memory
The plugin we built. Knowledge graph memory with librarian extraction, auto-recall, temporal invalidation. Available to the group.
See Config Guide →🛠️ Tools
Essential tools for working with AI agents.
Claude Code
Anthropic's coding assistant. SSH into servers, manage infrastructure, run tasks. The stepping stone to OpenClaw.
Docs →OpenAI Codex
Cloud-based coding agent with sandboxed execution. Parallel task processing in isolated containers.
openai.com →Telegram @BotFather
Create your Telegram bot — the primary interface for OpenClaw.
Open BotFather →Cloudflare
Tunnels for secure access, Pages for hosting, DNS management. Free tier covers most needs.
Dashboard →📚 Research & Sources
The engineering and research that informed this talk.
Cursor — Scaling Autonomous Coding
Wilson Lynn on the Planner → Worker → Judge architecture. Built a browser in Rust, then solved a math proof.
cursor.com/blog →Anthropic — Agentic Coding Trends 2026
How engineers delegate tasks they can "sniff check." Meta-skills over execution skills.
anthropic.com/research →Google DeepMind — AlphaProof
Separate generation, verification, and revision roles. The same pattern as peer review.
deepmind.google →The Convergence
Four organizations independently built the same architecture: decompose → parallelize → verify → iterate. The harness matters more than the model.
📄 The Illusion of Agent Learning
The Paper
The full breakdown — what your agent can and cannot do, the L1-L5 ladder, the semantic gap, and the three rules.
Read the paper →L1-L5 Capability Ladder
Your agent lives between L2 and L3. Here's why L4 and L5 don't exist yet.
View the ladder →The Semantic Gap
Two systems, one gap. The brain doesn't talk to the hands.
View the diagram →📋 Talk Guides
The handouts from this presentation.
Memory Configuration Guide
Every memory setting explained. Paste into your agent's context.
Open Guide →Telegram Commands Cheat Sheet
Essential commands + Telegram gotchas every user should know.
Open Cheat Sheet →Cognitive Memory Framework
The 10-layer chart mapping human memory to OpenClaw plugins.
View Framework →