🛠️ Tools demos

Step-by-step walkthroughs that show how to combine OpenClaw .NET's built-in tools into useful agent runs and scheduled jobs.

These demos are deliberately small, recipe-style guides — pick one, follow the steps, and you'll have a working scenario in 10–20 minutes. They focus on the wiring (which tools, in what order, with what prompt), not on building new code.

Looking for the full tool reference? See docs/manuals/20-tools.md. Looking for the bigger end-to-end scenarios? See real-world/.

🎯 Quick start: every demo on this page is also shipped as a built-in job template. Open the Web UI → Jobs → Job Templates to seed any of them with one click.

Demos in this track#

#	Demo	Surface	Tools used
01	Watched folder → markdown → summary (scheduled job, every 5 min)	Jobs page	`file_system`, `markdown_convert` (+ agent summarization)
02	GitHub issue triage (one-shot agent run)	Chat / Agent Probe	`github`, `html_query`
03	Research & archive (web → markdown → local semantic index)	Chat	`markdown_convert`, `file_system`, `embeddings`
04	Image batch resize (hourly job, folder → WebP thumbnails)	Jobs page	`file_system`, `image_edit`
05	Text → speech snippet (one chat turn produces a WAV)	Chat / Direct Invoke	`text_to_speech`

More demos will be added as new tools land. Suggestions and PRs welcome.

Prerequisites (all demos)#

OpenClaw .NET running locally (aspire start src\OpenClawNet.AppHost) — see 01-local-installation.md.
A model provider configured (Ollama, Azure OpenAI, GitHub Copilot, or Foundry).
For demo 1: a folder you can write into (we'll use c:\temp\sampleDocs).
For demo 2: optional GITHUB_TOKEN secret if you want to avoid GitHub's anonymous rate limit (see the demo for setup).

How these demos are organized#

Each subfolder contains:

README.md — the full step-by-step guide (UI-first, with a copy-paste API alternative at the end).
Optional *.json payloads or screenshots when they help.

You can read them in any order; they don't depend on each other.