hn.today

Show HN: DocuMemory – Persistent memory and context for AI coding tools

documemory.dev4 points0 comments
Screenshot of Show HN: DocuMemory – Persistent memory and context for AI coding tools

DocuMemory is a memory engine that turns AI coding chat sessions into persistent, structured engineering knowledge so future sessions start with the decisions and constraints that mattered rather than a full transcript. It passively captures conversations from clients like Claude Code, ChatGPT, and Cursor via a Model Context Protocol (MCP) integration, runs a "subtractive distillation" to extract validated facts and implementation reasoning, and produces compact context snippets ready to be injected back into prompts. The product emphasizes continuity for engineers by keeping successful solutions and their rationale while discarding failed attempts and noise, enabling faster onboarding of new sessions and team members.

Under the hood it stores distilled facts in a vector-backed Postgres setup (pgvector with HNSW indexing) and combines semantic and exact-match retrieval through a hybrid search pipeline using reciprocal rank fusion (RRF). The platform supports private, self-hosted deployments, encrypted storage, low-latency recall, and an API example for MCP configuration. Pricing begins with 15 free distillations, then $12/month (or $10/month billed annually) for individual use and $15/seat for teams. The offering targets engineers who want compounding, recallable design decisions and compact prompt real estate for AI-assisted development.

Read on documemory.dev0 comments on Hacker News

Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.

More in AI

The daily digest

Today's best Hacker News stories, summarized and screenshotted, one email a day.