Outcome-based persistent memory MCP server for Claude Code and OpenCode. Good advice promoted, bad advice demoted. pip install roampal.
Config is the same across clients — only the file and path differ.
{
"mcpServers": {
"roampal-core": {
"args": [
"roampal"
],
"command": "uvx"
}
}
}Are you the author?
Add this badge to your README to show your security score and help users find safe servers.
Works with Claude Code and OpenCode.
This server supports HTTP transport. Be the first to test it — help the community know if it works.
Five weighted categories — click any category to see the underlying evidence.
No known CVEs.
Checked roampal against OSV.dev.
Click any tool to inspect its schema.
Be the first to review
Have you used this server?
Share your experience — it helps other developers decide.
Sign in to write a review.
Others in ai-ml / productivity
Persistent memory using a knowledge graph
Privacy-first. MCP is the protocol for tool access. We're the virtualization layer for context.
Official Miro MCP server - Supports context to code and creating diagrams, docs, and data tables.
An open-source AI agent that brings the power of Gemini directly into your terminal.
MCP Security Weekly
Get CVE alerts and security updates for Roampal Core and similar servers.
Start a conversation
Ask a question, share a tip, or report an issue.
Sign in to join the discussion.
Outcome-Based Persistent Memory MCP Server
Two commands. Your AI coding assistant gets outcome-based memory.
Works with Claude Code and OpenCode.
85.8% on the corrected LoCoMo benchmark (non-adversarial, end-to-end answer accuracy) — validated on 1,986 questions across 10 conversations with dual grading. All figures in this section are sourced from the paper and roampal-labs (see citations at the bottom of this section).
| Result | Score |
|---|---|
| Conversational learning vs raw ingestion | +23 points (76.6% vs 53.0%, p<0.0001) |
| Architecture vs model effect | Architecture ~10x larger contributor |
| Poison resilience (1,135 adversarial memories) | -2.6 to -4.2 points only |
| TagCascade retrieval (tags-first + CE rerank) | +1.9 Hit@1 vs pure CE (p<0.0001) |
Benchmark pipeline runs on a single GPU with no cloud dependencies. Roampal itself runs on CPU — no GPU required. Full methodology, data, and evaluation scripts: roampal-labs
Paper: "Beyond Ingestion: What Conversational Memory Learning Reveals on a Corrected LoCoMo Benchmark" (Logan Teague, April 2026)
pip install roampal
roampal init
Auto-detects installed tools. Restart your editor and start chatting.
Target a specific tool:
roampal init --claude-codeorroampal init --opencode
The core loop is identical — both platforms inject context, capture exchanges, and score outcomes. The delivery mechanism differs:
| Claude Code | OpenCode | |
|---|---|---|
| Context injection | Hooks (stdout) | Plugin (system prompt) |
| Exchange capture | Stop hook | Plugin session.idle event |
| Scoring | Main LLM via score_memories tool | Independent sidecar (your chosen model, disabled by default until configured) |
| Self-healing | Hooks auto-restart server on failure | Plugin auto-restarts server on failure |
Claude Code prompts the main LLM to score each exchange via the score_memories tool. OpenCode never self-scores — an independent sidecar (a separate API call) reviews each exchange as a third party, removing self-assessment bias. The score_memories tool is not regis