1. Conduid
  2. RAG
  3. Memrain
MCP server · RAG

Memrain

by · github.com/timurgaleev/memrain

Self-hosted memory for your AI agents: hybrid vector + keyword + entity-graph search over your notes and code, with cited sources, served over MCP to Claude Code, Cursor and Codex. Runs in your own AWS account with team access controls, per-caller spend caps and zero telemetry. MIT.

Unclaimed ai-agentsai-memoryawsaws-bedrockbunclaude-codecode-searchcursorhybrid-searchknowledge-graphmcpmcp-servermodel-context-protocolpgvectorpostgressecond-brainself-hostedterraformtypescriptvector-search
InstallView source
39Low

Scored 13 days ago · breakdown

About Memrain

Memrain is an MCP server published by timurgaleev in the RAG category: self-hosted memory for your AI agents: hybrid vector + keyword + entity-graph search over your notes and code, with cited sources, served over MCP to Claude Code, Cursor and Codex. Runs in your own AWS account with team access controls, per-caller spend caps and zero telemetry. MIT.

The repository has 11 stars and 0 forks, with the last commit 7 days ago. Recent commit activity is the strongest signal that reported issues will get attention.

Install

From source
git clone https://github.com/timurgaleev/memrain

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Memrain

Powered by Claude · Grounded in docs

I know everything about Memrain. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

6 checks have not been run yet.

Releases

v1.2.0v1.2.0 · 10 Oct 2026Added Six ranking changes ship behind `MEMRAIN_RANK_*` flags, all off by default: `METADATA_GATE` (metadata boosts only when a lexical arm voted), `ENTITY_TRUTH_SOFT`, `KEYWORD_OR_RETRY` (OR keyword pass when the vector arm is empty),…
v1.1.0v1.1.0 · 10 Oct 2026Added `page_edit` changes part of a page: exact old/new text replacements at a given version, all or nothing, with facts and takes fences protected and a diff in the answer (redacted like any retrieval for non-operator callers); it refuses…
v1.0.9v1.0.9 · 10 Oct 2026Fixed Retrieval scores are measured per page. Several chunks of one page used to count as separate hits, which inflated recall and skewed MRR in `eval`, the replay probe and A/B compares. Queries with no expected answer no longer score a…
v1.0.8v1.0.8 · 9 Oct 2026Fixed The deployed stack now passes `MEMRAIN_UTILITY_MODEL` to the container, so setting it in `.env` moves every utility-tier call, not only the ones with a per-feature key. The agent loop can run on models that keep thinking on, such as…
v1.0.7v1.0.7 · 8 Oct 2026Added Memrain runs on the Claude 5.5 models. Point `MEMRAIN_UTILITY_MODEL` or `MEMRAIN_FACTS_MODEL` at `eu.anthropic.claude-haiku-5-5` (about a tenth of Haiku 4.5's price) or `eu.anthropic.claude-sonnet-5-5`. These models reject a custom…
Questions

About Memrain

How do I install Memrain?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Memrain safe to use with an AI agent?

Its trust score is 39 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Memrain still maintained?

Yes — the latest release is v1.2.0 (10 Oct 2026), and the last commit was 7 days ago. The repository has 11 stars.