Topics
Deep dives into how OpenClaw works under the hood. Pick a topic to explore.
System Prompt Assembly
How OpenClaw builds its system prompt from 26 sections — entry points, prompt modes, project context files, plugin hooks, and configuration knobs.
Memory Management
How OpenClaw persists and retrieves memories — two-layer storage, hybrid search pipeline, embedding providers, memory flush, and configuration.
Session Viewer
Debug timeline showing what the AI agent did step-by-step in each session — messages, tool calls, results, and model changes. Requires login.
Skill Viewer
Browse and inspect all installed OpenClaw skills — bundled and workspace. View documentation, requirements, and file structure.
Claude Code Viewer
Real-time dashboard for Claude Code tasks — monitor milestones, progress timeline, hook logs, and task output as coding agents work.
Tool Permission System
Deep dive into OpenClaw's multi-layer permission stack — tool profiles, allow/deny lists, exec security modes, safe-bin policy, approval lifecycle, shell hardening, and elevated execution.