Claw field notebook · v0b · multi-vendor
edit on GitHub colophon

· Pick a reading path

Three doors in. Each one leads to a section page where every entry carries a verification badge.

§ 01 I'm new to OpenClaw Five-minute orientation The 30-second version, the seven canonical workspace files, a flat schematic of the runtime, and the drawbacks the official docs won't lead with. Open §1 Overview → § 02 I want to install Pick a setup path Six install paths from laptop to Pi to Azure, plus a hardware matrix. Concrete commands. Honest verification states on each page. Open §2 Setup → § 07 I'm comparing stacks OpenClaw vs the alternatives Honest matchups against MCP-based stacks, Claude Desktop + plugins, hosted SaaS. Where OpenClaw fits — and where it doesn't. Open §7 Compare →

· Browse by section

All ten sections. Click any tile to open the section's catalog.

§ 1 Overview 4 Five-minute orientation. Concepts, workspace files, schematic. § 2 Setup 6 Install paths from laptop to Pi to Azure, plus a hardware matrix. § 3 Connections 5 Channels, tools, models, memory — the integration patterns. § 4 Plugins 1 Field notes on marketplace plugins. Sourced, not endorsed. § 5 Use cases 5 End-to-end recipes. Hello-world, Slack agent, RAG, Pi home. § 6 Security 4 Patterns to do (and not do) when self-hosting an agent runtime. § 7 Compare 6 OpenClaw vs MCP-based stacks. Honest matchups, no marketing. § 8 Updates 11 Hand-curated. Not a CVE feed. Date · type · source link. § 9 Resources 12 The good stuff written elsewhere. Linked, not duplicated. § 10 FAQ 8 Real questions Sush had. Production readiness, cost, longevity.

· What's new

Hand-curated. Not a CVE feed. See the full feed →

2026-05-15 NEWS
Claw Planet · Batch E — Compare + Updates seed shipped

Three cross-vendor /compare/ pages live — CLI coding agents (Claude Code · Codex CLI · Gemini CLI · GitHub Copilot CLI), direct model APIs (Anthropic · Gemini · Foundry), and M365 extensibility paths (Copilot Studio · Declarative Agents · Custom Engine Agents). The comparison shape is consistent across all four entries — where each wins, where each lags, honest take at the bottom. Phase 1 of v0b complete.

2026-05-14 RELEASE
Claude Code 2.1.142 — claude agents flags · Fast-mode Opus 4.7

Anthropic's `@anthropic-ai/claude-code` 2.1.142 (npm registry, 14 May 2026; on the `latest`/`next` channel — `stable` lags at 2.1.132) adds eight new `claude agents` flags (--add-dir · --settings · --mcp-config · --plugin-dir · --permission-mode · --model · --effort · --dangerously-skip-permissions) for configuring dispatched background sessions. Fast mode now defaults to Claude Opus 4.7 (was Opus 4.6); pin to 4.6 with `CLAUDE_CODE_OPUS_4_6_FAST_MODE_OVERRIDE=1`. Note: npm install is deprecated per the current README — use `curl -fsSL https://claude.ai/install.sh | bash`.

2026-05-12 RELEASE
Gemini CLI 0.42.0 — Gemma 4 enabled by default

Google Gemini CLI 0.42.0 (npm registry, 12 May 2026) enabled Gemma 4 models by default for all users via the Gemini API. Also adds Auto Memory Inbox (a canonical-patch contract for seamless skill management) and Voice Mode wave animations with privacy/compliance UX warnings for the Gemini Live backend. Free tier: 1,000 req/day on a personal Google OAuth account (60 req/min cap); same daily cap via an AI Studio API key with explicit model-selection control.