OC-301e · Module 2

Plugin Discovery & Recommendation

3 min read

A registry with 500 plugins is only useful if the user can find the one they need. Discovery mechanisms surface the right plugin based on the user's context: their current workflow, their agent configuration, and the gap they are trying to fill. Three discovery mechanisms work in combination.

Search: keyword and category search across plugin names, descriptions, and tags. This handles the case where the user knows what they are looking for. Recommendation: based on the user's installed plugins and usage patterns, recommend plugins that are commonly installed alongside them. "Users who installed the CRM connector also installed the email template plugin." Gap analysis: based on the agent's current capabilities and the tasks it frequently encounters, identify plugins that would fill a capability gap. "Your agent handles data extraction but has no visualization plugin — consider installing the chart generator."

Do This

  • Provide multiple discovery paths: search for known needs, recommendations for related capabilities, gap analysis for unknown needs
  • Surface usage statistics and ratings prominently — social proof helps users evaluate quality
  • Recommend based on the user's actual configuration, not generic popularity — relevance beats popularity

Avoid This

  • Rely on search alone — users cannot search for capabilities they do not know exist
  • Rank plugins only by download count — a popular plugin for a different use case is not relevant
  • Hide compatibility information until after installation — save the user time by showing compatibility upfront