AI Tags seems broken

[tcsenpai]

Describe the bug
Generating AI tags result in a sentence (often multi line) instead of just tags.
Additionally, tags are then introduced within markdown contents resulting in H1 titles instead of actual tags.

To Reproduce
Steps to reproduce the behavior:
Generate AI tags for a Blinko

Expected behavior
Get a list of tags

Screenshots

Desktop (please complete the following information):

• OS: Linux
• Browser: Zen (Firefox)
• Version latest

Additional context
Using default AI settings with Ollama and llama3.2

[dosubot]

Hey there, @tcsenpai! I'm here to help you with any bugs, questions, or contributions. Let's tackle this together!

The issue you're experiencing with AI tag generation in Blinko, where the output is a sentence instead of discrete tags, might be related to how the api.ai.autoTag.mutate function processes the content. Unfortunately, the specific logic used by this function to generate tags isn't detailed in the available information, so it might require further investigation into the implementation of this function.

Additionally, the problem with tags being inserted as H1 titles in markdown content could be due to the way markdown is handled when inserting tags. The replaceMarkdownTag method is responsible for replacing markdown content with new tags, and it formats the tags as "#" + text + ' ' ^[1]. This might be causing the tags to be interpreted as headers if the markdown syntax isn't correctly managed.

To address these issues, you might want to:

1. Review the implementation of the api.ai.autoTag.mutate function to ensure it outputs tags in the expected format.
2. Check the replaceMarkdownTag method to ensure it correctly formats tags for markdown content without converting them into headers.

These steps could help in resolving the issues with AI tag generation and insertion.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Other}

[tcsenpai]

Great to have a bot replying, just bumping this to ensure an human will actually read it

[blinko-space]

Optimized the prompt. It should be more stable now.