AgentLimb · Onboard Prompt is part of Hooosberg's public product diary for AgentLimb. It turns a real build step into a searchable case study for international readers.
AgentLimb is a Chrome extension plus a local MCP host for universal browser control. Any MCP-capable agent — Claude Code, Codex, Cursor, Tencent WorkBuddy — pairs by copying one prompt with automatic authorization by default, drives multiple Chrome profiles with targeted routing, supports native HTTP Tool RPC calls, and reuses local Markdown site memory (~/.agentlimb/memory) — 100% local.
The original Chinese article was published on 2026-06-11 under AI workflow. It focuses on 工具越底层，接入越要简单；AgentLimb 把配置步骤收敛成可复制的提示词。
The reusable learning angle is Browser automation, MCP toolchains, Site memory playbooks, AI operating real websites. The diary format is deliberately practical: context, decision, implementation boundary, review impact, and what another builder can reuse.
This English page is generated from the same public product facts as the Chinese site. It is meant to be indexable and useful now, while still leaving room for later hand-polished translation of the full Chinese narrative.
Context
This entry belongs to the AgentLimb build series. It should be read as one stage in a larger product journey, not as an isolated announcement.
The site keeps success cases and failure cases together because both create reusable judgment.
Implementation boundary
Relevant product capabilities include One copied prompt to connect (auto-authorized by default), Multi-account Chrome profile routing, Native HTTP Tool RPC (call tools without stdio), Smooth multi-step lifecycle (anti-flicker tab marker), Pluggable local site memory (learn once, reuse across AIs), The side panel is the AI's reply window. The diary explains how these capabilities are selected, constrained, reviewed, or turned into launch material.
Privacy and trust
The extension and the local host run entirely on your machine. The MCP host listens only on 127.0.0.1, pairing trust and chat records live in the extension's private storage, site memory stores plain Markdown playbooks with automatic credential redaction, and page data is never uploaded — no cloud, no tracking.
For an indie product matrix, trust is built through many small facts: local data boundaries, review status, support links, honest postmortems, and stable canonical pages.
What another builder can reuse
Another builder can reuse the pattern: start with a narrow product promise, make the privacy boundary inspectable, connect every public page to a real download, support path, or source repository when appropriate, and turn the launch process into a searchable article.
That is why the diary sits next to the product page. The product builds trust, and the diary teaches the path that made the product possible.