Rudimentary API: Try using it and tell me what you think

General
43

@hugows @chiccoach

The <time> and <code> tags should now behave as before.
In addition, you can now submit markdown, and it will be converted into Workflowy’s formatting automatically, it’s not documented yet, but feel free to try it out and share feedback.

2 Likes
44

Hi guys,

I’ve just started experimenting with the API and wanted to share a bit of feedback. So far, I’ve run into a couple of difficulties:

  1. Do you plan to add a retrieve feature that accepts a string as input (instead of just the node ID)?
  2. Besides inspecting the HTML, is there a better or recommended way to get the ID of a node? (it’s linked in a kind of way the first question)
  3. Is there a proper way to store Markdown content through the API?

Thanks a lot for the api ! I will share more feedback overtime following my experimentations.

45

Confirmed

image
image836×320 42.3 KB

46

Asked Claude Code to code (one-shot) a Workflowy interactive browser in Python - it was pretty useful for this.

47

Dear @anatoliy,

I am really enjoying trying out the API, but one thing is a bit annoying for what I am planning:

When I am generating a new node with this endpoint:
curl -s -X POST https://workflowy.com/api/v1/nodes/ \

I would like to include have a backlink to another node. Actually, it is working fine, and the new node shows up where it should be with the working backlink.
BUT the respective node which is backlinked doesn’t show the new node under its backlink list. So somehow this action of generating an updated “backlink list” is not triggered via the API. When I do a manual change of the new node (e.g., deleting any letter and putting the letter back), the backlink list is updated.

Is there a way that backlinking is triggered via the API?

Thanks for your help.

Cheers,
Jan

1 Like
48

Same problem for me!!
See this message

@anatoliy

49

Is there a way to send something to today instead of a fixed parent like Inbox?

2 Likes
50

Hi, how can i get access to this slack usergroup?
cheers, Jan

51

Not sure to get it

52
53

My use case is using the api with LLMs. This means the LLM rarely has any context for where anything is each time I use it and so it has to make a lot of calls to make anything happen.

I’ve seen similar things requested in other threads but I’m resurfacing them here to keep these centralized.

These two endpoints I suggest in particular seem like they would be really useful in terms of making what I’m trying to do feasible. Right now it’s doable but not really practical unless you use some memory mechanism to avoid having the LLM from making a ton of calls.

  • Get all children up to n depth - This would allow me to make far fewer calls when extracting more than a handful of nodes at a time. Ideally we could set a max depth for browsing and no max depth to extract all children at any depth.

  • A search endpoint - would significantly reduce the number of calls I need to make. Unless I’m using some sort of memory, the LLM has no context for where anything is or the structure of the account so it’s going to make a ton of calls just to get its bearings. A single search call followed by an extract at n depth would replace many many calls to list nodes. If something is ten levels down, that’s ten calls vs two. And that’s assuming I’m going to make all the changes in that same location ten levels down, if I want it to find something ten levels down and have it affect something somewhere else that’s also ten levels down - I think you get the idea.

54

Hi Rodolfo,

I recently built a CLI, which implements retrieving nodes recursively, up to a configurable depth as well as a search capability, searching within all nodes (using the Retrieve and ExportAll endpoints).

I also just recently added access to these through an MCP server, so that you can issue commands from your favorite LLM.

I wrote a blog post about using it with Claude:

I’m always happy to get feedback, so let me know what you think.

Thanks

Marc

4 Likes
55

Just adding some more requests around mirror support.

Basics

  • Ability to get the mirrors via the api
  • Some sort of indicator that the item is a mirror

Wishlist

  • Ability to create mirrors
  • Ability to detach mirrors
  • Ability to get all other mirrors of a mirror
3 Likes