Docs

Author: robert7

docs/demo.md

@@ -2,11 +2,51 @@
 
 Demonstration of `remnote-cli` against a live RemNote + RemNote Automation Bridge plugin setup.
 
+## Use RemNote from Any Coding Harness
+
+No MCP server config needed. In any coding harness — Claude Code, GitHub Copilot CLI, Codex CLI, or similar — you
+can interact with RemNote by pasting a single prompt that loads the skill and drives the CLI.
+
+**Prerequisites:** `remnote-cli` installed globally, RemNote Automation Bridge plugin running in RemNote.
+
+### How it works
+
+1. Paste the prompt below into your coding harness.
+2. The harness fetches
+   [SKILL.md](https://github.com/robert7/remnote-cli/blob/main/skills/remnote/SKILL.md) — a concise guide that tells
+   the agent how to use `remnote-cli`.
+3. The agent runs CLI commands against your local RemNote bridge.
+
+No config files, no MCP server setup.
+
+### Example prompt
+
+```text
+read https://github.com/robert7/remnote-cli/blob/main/skills/remnote/SKILL.md
+
+check remnote status using remnote CLI
+search remnote for "AI assisted coding"
+```
+
+### Claude Code
+
+![remnote-cli from Claude Code](images/remnote-cli-coding-harness-claude-code.jpg)
+
+Claude Code fetches SKILL.md via an agent subcommand, verifies daemon and bridge status, then searches RemNote — all
+in one turn, no prior MCP configuration.
+
+### GitHub Copilot CLI
+
+![remnote-cli from GitHub Copilot CLI](images/remnote-cli-coding-harness-copilot-cli.jpg)
+
+Same prompt in GitHub Copilot CLI. It fetches SKILL.md via the GitHub MCP tool, checks daemon and bridge status, and
+runs the search. The same approach works in VS Code Copilot and other harnesses that can read URLs.
+
 ## Example of OpenClaw Chat Workflow with `remnote-cli`
 
 ![remnote-cli with OpenClaw](images/remnote-cli-openclaw-1.jpg)
 
-This screenshot shows [OpenClaw](https://github.com/openclaw/openclaw) driving 
+This screenshot shows [OpenClaw](https://github.com/openclaw/openclaw) driving
 `remnote-cli` from a WhatsApp chat: it checks bridge connectivity, runs a
 `search` for "blue light", returns matching note IDs, and then attempts a `read`-based summary. The flow highlights
 agent-to-CLI orchestration, quick status verification, and transparent handling when note content is incomplete.