Quickstart (10 Minutes)
Verifiable cognitive memory for personal vaults. Cite-or-it-didn't-happen.
Memora retrieves claims, not notes - atomic facts with source-span pointers, validity windows, and privacy bands. Every LLM citation is architecturally validated against your markdown.
This walkthrough takes a determined user from install to first verified cited answer in about 10 minutes.
1) Install Memora
curl --proto '=https' --tlsv1.2 -LsSf \
https://github.com/radotsvetkov/memora/releases/latest/download/memora-cli-installer.sh | sh
What just happened
The installer places the Memora binaries on your system path (typically including
memora and memora-mcp), so you can run CLI commands and expose MCP tools to
clients without manual build steps.
2) Initialize a vault
memora init --vault ~/brain
What just happened
Memora created a vault scaffold with:
world_map.md(generated memory overview surface)- a sample semantic region folder
.memora/config.tomlfor local configuration and sidecar state
Your markdown vault stays human-readable and Obsidian-compatible from the start.
3) Add your first note manually
Create ~/brain/semantic/projects/drift/roadmap.md:
---
title: drift Serialization Decision
type: semantic
date: 2025-09-12
region: projects/drift
privacy: private
tags: [drift, serialization, architecture]
---
drift switched its serialization format from JSON to MessagePack in Q3.
Benchmarking showed a 3x throughput improvement.
Decision recorded in roadmap, retro, and review notes.
What just happened
You wrote plain markdown with frontmatter. No proprietary format is required. Memora will use this as extraction input and anchor future citations to precise source spans inside this file.
4) Index the vault
memora index --vault ~/brain
What just happened
Indexing runs a full pipeline:
- note metadata parsed and recorded
- claim extraction generates atomic claim candidates
- each stored claim is linked to source byte range and fingerprinted
- lexical and vector indexes are updated
- contradiction checks compare new/current claims and supersede older conflicts
The result is a queryable claim graph rather than a plain chunk index.
5) Ask your first query
memora query "What did we decide about drift's serialization format?" --vault ~/brain
What just happened
Memora retrieved claims, assembled model context, generated an answer, and validated cited claim ids against source spans before returning output. In the answer footer, citation entries correspond to claim ids that survived validation.
6) Add Claude Code MCP integration
Add one server entry to your MCP config:
{"mcpServers":{"memora":{"command":"/usr/local/bin/memora-mcp","env":{"MEMORA_VAULT":"/absolute/path/to/brain"}}}}
What just happened
Claude Code can now call Memora tools over stdio MCP. Instead of passing raw note chunks manually, you get memory-tool access with claim-aware retrieval and built-in citation validation behavior.
7) Ask the same question in Claude Code
Prompt:
What did we decide about drift's serialization format?
What just happened
Claude Code routed the request through Memora MCP tools. The response surfaced with verified citations tied to your markdown-backed claims. If a citation fails validation, Memora strips/retries with verified-only context before returning.
Next 5 minutes (recommended)
- Run
memora watch --vault ~/brainin a dedicated terminal tab. - Add inline secret markers in mixed-sensitivity notes.
- Check
world_map.mdafter your next capture session. - Use
memora_record_usefulwhen a cited answer was especially helpful.
Troubleshooting basics
command not found: memora-> restart shell or verify installer path export.- No claims extracted -> confirm frontmatter + note content + successful indexing.
- MCP tool unavailable -> verify
memora-mcppath andMEMORA_VAULTvalue. - Sparse answers -> add more factual statements; reflection-only notes yield fewer claims by design.