Skip to content

memory Cog

Captures Discord content verbatim, tags it, and prepares it for retrieval through the ChainStore interface (planned qdrant backend).

Roadmap highlights

  • MVP: commands to remember messages, light grammar cleanup, tagging, and vectorization through ChainStore once available.
  • Retrieval: [p]memory search <query> and ChainHub tools (memory_search, memory_save) to surface memories during conversations.
  • Context injection: optionally enrich agent replies with relevant memories.
  • Advanced plans: optional summarization, conversation-wide summaries, embed-driven UI, and batch operations for import/export or re-embedding.
  • Dependencies: requires an operational ChainStore backend; inspired by vertyco/vrt-cogs/assistant.