Reference MCP vs RoBrain

Reference MCP and RoBrain are both agent frameworks tracked by AIDiveForge. Below is a side-by-side comparison of pricing, capabilities, platforms, and ownership — sourced from each tool's live website and verified before publishing.

Reference MCP

Reference is a local MCP server that reads session transcripts and memory files — CLAUDE.md, AGENTS.md, and similar — from multiple AI coding tools, then exposes that history to whichever agent is asking. Register it once in each tool's MCP config and any agent can query what any other agent did before. The architecture is passive: Reference does not plan, execute, or chain tasks — it answers lookups. That scope is the point, and also the ceiling. Teams with more than a handful of tools, or who need structured, queryable memory rather than transcript search, will find the file-scanning approach starts to feel fragile as session volume grows.

RoBrain

RoBrain sits between your team's AI coding tools — Claude Code, Cursor, Copilot, Codex CLI — and a shared Postgres instance, capturing not just decisions but the alternatives your team ruled out. An MCP server runs inside the editor and surfaces relevant history before the agent acts; a batch Synthesis scan reads the whole corpus on a schedule to flag contradictions and drift that no single session would catch. That cross-session contradiction detection is where it separates from alternatives that only check at insertion time or silently delete the losing decision. Self-hosted on Apache 2.0 with your own Postgres; cloud extraction and the Planning API are paid-only features.

Attribute	Reference MCP	RoBrain
Pricing	Free	Paid
Free trial	No	No
Open source	Yes	Yes
Has API	No	Yes
Self-hosted option	Yes	Yes
Platforms	Python / local	Node.js 18.18+, npm/pnpm; Docker for local Postgres + Perception API; integrates with Claude Code, Cursor, Copilot, Codex CLI
Released	—	2026
Pros	Reads session transcripts from Claude Code, Codex CLI, and Cursor in one place, so you stop re-explaining decisions that were already made in a different tool last week. Fully local and offline, which means code and conversation history never leave your machine — critical for projects where sending context to a third-party sync service is off the table. MIT-licensed with no paid tiers and no account required, so there is no vendor relationship to manage and no access cliff if a pricing tier changes. Single MCP registration pattern works across supported tools, so you configure it once per tool rather than wiring a custom integration for each pair.	Stores rejected alternatives as a structured field alongside each decision, so the agent surfaces why an approach was ruled out — not just what was chosen — before it re-proposes something your team already vetoed. Cross-session Synthesis scan reads the entire decision corpus on a schedule, so contradictions that accumulate across weeks and multiple developers get flagged rather than sitting invisible until they cause a revert. Old and new decisions both stay queryable when your team changes course, so reconstructing why a reversal happened is a query, not a memory exercise — unlike tools that silently replace the losing decision. One shared Postgres for the whole team works across Claude Code, Cursor, Copilot, and Codex CLI simultaneously, so a decision made in one editor is visible to an agent running in another without manual sync. Apache 2.0 self-hosted path keeps decision history on infrastructure you control, so teams with data residency requirements or cost sensitivity on API calls can run the full open-source layer without a cloud dependency.
Cons	Search runs over raw transcript files and markdown — not a structured index. Once session volume is high enough that the answer you need is buried in dozens of transcripts, recall becomes a best-effort grep rather than a reliable query, and teams at that point look at purpose-built vector stores like a local Chroma or Weaviate instance instead. Memory is local to one machine. A team of two engineers running the same agents on different laptops gets zero shared context — Reference has no sync layer. Teams who need shared agent memory across contributors abandon this and wire a shared database backend, at which point Reference's architecture no longer fits. There is no API and no programmatic query surface outside the MCP protocol. Any workflow that needs to pull agent history into a dashboard, a CI pipeline, or a custom tool has no supported path — the docs describe no extension point for that use case.	The automatic pre-action warning — surfacing veto context before an agent makes an unsafe suggestion — is a cloud-only feature; self-hosted teams trigger inject queries manually via the CLI, which means the protection only fires when a developer remembers to ask for it, not automatically at the moment of risk. Synthesis runs as a scheduled batch scan, not in real time; a contradiction introduced between scans will not be flagged until the next run, so teams moving fast in a single day can still ship a conflicting decision before the corpus-wide check catches it. The value scales with history depth and team size — the vendor's own qualifier is that a project under a few months old with a single developer and one AI tool does not justify the setup cost. Teams in that situation who set this up and find the overhead exceeds the benefit tend to revert to a maintained CLAUDE.md and revisit RoBrain only when the codebase and team grow.

Bottom line

Reference MCP is free while RoBrain is paid; only RoBrain exposes a public API. Choose based on which difference matters most for your workflow.

Frequently asked questions

What is the difference between Reference MCP and RoBrain?

Reference MCP is Free and open source, while RoBrain is Paid and open source. Compare pricing, free trial, API, platforms, and pros/cons in the table above on AIDiveForge.

Is Reference MCP better than RoBrain?

It depends on your workflow. Use the side-by-side attributes (pricing, open source, API, self-hosted, platforms) to decide. AIDiveForge does not rank a universal winner — we publish verified facts so you can choose.