Predict context rot, sign every agent write, tier reads. Markdown-native, 100% local.
MCP server exposing Mnemoscope's context-rot scoring, tiered reading, and signed journal tools to any Model Context Protocol–compatible client (Claude Code, Cursor, ChatGPT desktop, anything that speaks MCP over stdio). 100% local. No cloud. No telemetry. Apache-2.0. Or run via without a global install: Restart Claude Code. The four tools below are discovered automatically. Tool | Input | What it…
Verification confirms publisher identity (repo ownership), not code safety. The security scan covers known CVEs and suspicious install scripts — it cannot prove the absence of malicious code.
MCP server exposing Mnemoscope's context-rot scoring, tiered reading, and signed journal tools to any Model Context Protocol–compatible client (Claude Code, Cursor, ChatGPT desktop, anything that speaks MCP over stdio). 100% local. No cloud. No telemetry. Apache-2.0. Or run via without a global install: Restart Claude Code. The four tools below are discovered automatically. Tool | Input | What it returns | | | 0–100 risk score, dominant factor, full factor breakdown, top-5 risk notes, vault…