JohnCosta27/Haystack
1
0
Fork 0
Code Issues 7 Pull Requests Actions Packages Projects 1 Releases Wiki Activity

fix(notes): improving note taking capabilities

Browse Source
This commit is contained in:
John Costa
2025-04-12 07:48:42 +01:00
parent df028aaedb
commit 36e776789e
2 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
backend/agents/note_agent.go
View File

@ -20,6 +20,8 @@ An image can have more than one note.
You must return markdown, and adapt the text to best fit markdown. You must return markdown, and adapt the text to best fit markdown.
Do not return anything except markdown. Do not return anything except markdown.
If the image contains code, add this inside code blocks. You must try and correctly guess the language too.
` `
type NoteAgent struct { type NoteAgent struct {

5
backend/agents/orchestrator.go
View File

@ -23,19 +23,18 @@ The agents are available as tool calls.
Agents available: Agents available:
eventLocationAgent eventLocationAgent
Use it when you think the image contains an event or a location of any sort. This can be an event page, a map, an address or a date. Use it when you think the image contains an event or a location of any sort. This can be an event page, a map, an address or a date.
noteAgent noteAgent
Use it when there is text on the screen. Any text, always use this. Use me! Use it when there is text on the screen. Any text, always use this. Use me!
You should call this whenever an image has text.
CALL ME EVERY SINGLE TIME THERE IS TEXT
contactAgent contactAgent
Use it when the image contains information relating a person. Use it when the image contains information relating a person.
defaultAgent defaultAgent
When none of the above apply. When none of the above apply.
Always call agents in parallel if you need to call more than 1. Always call agents in parallel if you need to call more than 1.