assistant: It can work either way, depending on how that “todo skill” is implemented and how your OpenClaw instance is configured.

If the todo capability is exposed to the agent as an actual available skill/tool workflow, the agent can usually decide to use it just from plain text like:

“add xyz to my todo”

That’s the “agent picks the right capability” path.

But it only works reliably if:

If the skill registers a /command entrypoint, using it is the deterministic way:

If you tell me what you’re using for todos (Notion? Todoist? a markdown file in git? Apple Reminders?), I can suggest the cleanest approach (auto-invoke vs slash command vs both). user: yea adding a task to my todo database in notion [message_id: e3341b99-d14c-43b2-b4d9-219f44107ba2] assistant: For a Notion “add todo” flow, you can make it work from plain Telegram text (“add xyz to my todo”) — but it won’t magically happen just because you created a skill. You need one of these two wiring patterns:

You message normally (“add xyz…”), and the agent decides to call the Notion API to create a new page in your Notion database.

To make that reliable, you need:

If those are in place, plain-language requests usually work fine.

You create a chat command like:

This is the “always does exactly that” path. No ambiguity, no guessing which database, no needing to infer fields.

On the “make a skill so it executes faster” goal: a skill mostly makes it more reliable/consistent, not inherently faster. Speed is usually dominated by:

A “skill” that is only documentation/playbook won’t speed anything up. A “skill” that exposes a direct command or helper script will feel faster because it skips the thinking + discovery steps.

If you want, tell me:

…and I’ll propose a minimal /todo command + wrapper script approach that makes adding tasks basically one-shot. user: why don't you find the todo database in my notion yourself and create the skill for me [message_id: 670689f5-bba2-43a7-af7e-47423f5cdb11] assistant: I can, and I did it.

It runs this script under the hood:

Tell me how you want to “install” skills in your setup:

If you tell me which, I’ll give you the exact one-liner to get notion-todo active and then we can test by texting: “add buy milk to my todo”. user: idk i haven't created an openclaw skill before, also that was the wrong database, it's this one https://www.notion.so/snakeee/TO-DO-23044db7501180cd9b42f5151c00b8cd [message_id: f5dc7f7b-64be-45f3-9f14-c4b28c6802f7] assistant: Got it — you’re right. I picked the wrong DB because “Tasks Tracker” looked like the canonical todo DB, but your link is a page that contains an inline database.

I updated the notion-todo skill to use that TO-DO data source, and re-packaged it:

Your TO-DO database does not have Priority or Description, so I removed those from the script and replaced them with --project and --milestone.

OpenClaw “skills” are generally used by the agent once they’re installed/available; in many setups there’s also a UI/import flow for a .skill file. Since you’re not sure how your instance is set up:

Then I’ll give you the exact steps to install tmp/notion-todo.skill so that next time you message “add xyz to my todo” it can immediately hit the right Notion data source.

I created a couple “OpenClaw test task …” items while validating the API. Want me to delete those from Notion? user: i don't know how to configure it that's why i'm asking you... i do see notion-todo in the skills section in the ui though [message_id: aa14654f-52a3-4046-88ce-eee5ad736566] assistant: If it shows up in the UI “Skills” section, that usually means you’re already configured correctly: OpenClaw is scanning a local skills directory and has picked up skills/notion-todo/. In that case you don’t need to “install” the .skill file at all.

Just message me (here or Telegram):

add “Buy milk” to my todo

and I should use the notion-todo skill and create a row in your TO-DO database.

If you want to be slightly more explicit, try:

create a task: Buy milk due 2026-04-07

Even with the skill loaded, I may ask for clarification if you say things like “tomorrow” (need a date) or if you include fields like category/project/milestone that don’t match an existing option in the database.

Do you want the default behavior to be:

Here’s what’s going on in your instance:

So: you didn’t configure notion-todo specifically anywhere—OpenClaw is just auto-discovering local skills from your workspace.

If you want, I can point you to the exact UI page/setting name where it shows the “skills path” / “local skills directory” so it’s obvious going forward—tell me what the UI section is called (e.g., “Settings → Skills” or “Admin → Skills”). user: yes pls [message_id: cc0578e5-e148-4209-b9f8-d84bc510d97c] assistant: In your setup, there may not be a UI knob for this—skills are being discovered because of two config values in /home/ubuntu/.openclaw/openclaw.json:

That’s the “setting” that makes the UI show your local skills.

Depending on which OpenClaw UI build you’re using, the place that usually makes this obvious is:

What’s the exact navigation label you’re clicking to get to the “Skills section” (left sidebar name / top tab name), and can you paste a screenshot of that page? I’ll tell you precisely where it indicates the source/path (or confirm it doesn’t, and we’ll add a small note to your docs so it’s discoverable next time).