Feature request: Let Capacities AI read and use tag descriptions

Hi Capacities team — I’d love to share a small feature request that I think would make AI auto-tagging and “Fill properties” much more accurate for people who use structured tag systems.

Right now, it’s not clear whether the AI takes tag descriptions into account when it decides which tags to apply (either via Auto-tagging or when filling tag-based properties (e.g., for me: topic/, course/, task/, source/, etc.). I maintain detailed tag descriptions specifically to capture the meaning and rules for when a tag should be used, and I’d love for the AI to benefit from that to increase the accuracy of its decisions when I click on fill properties.

Why this would help

From my experience, I've found that in larger tag libraries, tag names alone often aren’t enough to disambiguate intent. Descriptions usually contain the “semantic contract” of a tag (what it means, what it’s for, and what it’s not for). If the AI can’t see those descriptions, it’s more likely to:

• pick the wrong tag among similar options
• over-tag or miss important tags
• blur boundaries between families (e.g., using a

What I’m hoping for

When the AI is filling a tag property, would it be possible to please include the descriptions of candidate tags (and optionally the tag collection description/rules) in the context sent to the model.

A few ways this could work smoothly:

• Only include tag descriptions for the relevant family (e.g.,
• If performance is a concern with very large tag sets, include descriptions for a shortlisted subset (top N candidates), or cap the total text length

Optional (but would be amazing)

• A setting like: “Use tag descriptions in AI context: On/Off”
• A small “what the AI used” preview/debug view so users can confirm descriptions were included

Thanks for considering this, I think it would meaningfully improve accuracy for anyone like myself building a more structured second brain system in Capacities. Happy to share screenshots/examples if helpful.