Couchloop EQ

Name: Couchloop EQ
Author: wisenbergg

@wisenbergg

Couchloop EQ について

CouchLoop EQ is an MCP (Model Context Protocol) server that provides behavioral governance for LLMs. It monitors AI responses for hallucination, inconsistency, tone drift, and unsafe reasoning patterns, while also managing stateful sessions and guided journeys that remember where

#ai-safety #developer-tools #code-review #security #session-management

基本情報

カテゴリ

その他

トランスポート

stdio

公開者

wisenbergg

投稿者

Greg

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "couchloop-eq": {
      "url": "https://mcp.couchloop.com/mcp",
      "transport": "streamable-http"
    }
  }
}

ツール

Start a new guided session. Optionally specify a journey to follow.

Send a message through the therapeutic AI stack with crisis detection and emotional support.

Resume a previously paused session. Returns current state and next step.

Save progress or capture a key moment in the current session.

Get all checkpoints for a session.

List available guided journeys/experiences.

Get current progress in a session/journey.

Capture a meaningful insight or realization from the conversation.

Get user insights, optionally filtered by session.

Get relevant context about this user for personalization.

Validate package dependencies to prevent installation of non-existent or malicious packages

Pre-screen AI-generated code to catch obvious issues and reduce human review burden. Detects console logs, commented code, TODOs, missing error handling, hardcoded values, type issues, and code complexity problems.

Comprehensive security scanner for AI-generated code. Detects SQL injection, XSS, hardcoded secrets, and other vulnerabilities with CWE codes and secure alternatives.

Detect code bloat, verbose patterns, and over-engineered code that creates technical debt. Analyzes complexity metrics and suggests simplifications.

Prevents accidental file deletion and destructive operations. Validates operations against protected paths and patterns, creates automatic backups, and provides rollback capability.

Get current file protection status, configuration, and backup statistics.

Get history of file operations that were attempted or executed.

List all available backup snapshots that can be restored.

Restore a file from a backup snapshot (undo a file modification/deletion).

Enable code freeze mode - all file operations require explicit approval (critical protection mode).

Disable code freeze mode - resume normal file operation protection rules.

Store, retrieve, and manage important project context to prevent AI amnesia when context windows fill up. Preserves architectural decisions, requirements, constraints, and technical patterns across conversations.

Validate library versions against latest releases, detect deprecated APIs, and prevent AI from using outdated patterns. Checks for breaking changes, security vulnerabilities, and provides upgrade paths.

概要

What is Couchloop EQ?

Couchloop EQ is a behavioral governance layer for AI assistants built on the Model Context Protocol (MCP). It monitors for hallucination, inconsistency, and unsafe reasoning patterns while managing stateful AI sessions, helping developers and users maintain safer, more consistent conversations.

How to use Couchloop EQ?

Couchloop EQ works with any MCP‑compatible client (Claude Desktop, ChatGPT, Cursor, Windsurf, VS Code, etc.). The easiest way is to connect to the hosted production endpoint at https://mcp.couchloop.com/mcp using streamable-http transport. For local use, install the npm package couchloop-eq-mcp globally and add the server configuration to your client’s MCP settings. No signup or API key is required; sessions persist automatically to ~/.couchloop-mcp/identity.json.

Key features of Couchloop EQ

Behavioral governance with hallucination detection and consistency checking
Stateful session management preserving context across interactions
Privacy‑by‑design: no personal data stored, session‑based isolation
Ten primary MCP tools: couchloop, guard, verify, status, conversation, brainstorm, code_review, package_audit, remember, protect
Safety guardrails against harmful advice, clinical overreach, and manipulation
OpenTelemetry distributed tracing across all stages

Use cases of Couchloop EQ

Catch hallucinated npm, PyPI, or Maven packages before installation
Detect insecure code (SQLi, XSS, hardcoded secrets) via automated code review
Save and recall architecture decisions across sessions using remember
Protect files from accidental deletion with backup, freeze, and rollback
Guide structured brainstorming for trade‑off analysis and feature design

FAQ from Couchloop EQ

What clients does Couchloop EQ support?

Any MCP‑compatible client, including Claude Desktop (v0.7.0+), ChatGPT (via Developer Mode), Cursor, Windsurf, VS Code, and others.

Does Couchloop EQ require authentication or an API key?

No, no authentication or API key is needed. Sessions are isolated by session ID, and no personal data is stored.

Where is session data stored?

When using the hosted server, data is managed server‑side with session isolation. For local installations, sessions persist automatically to ~/.couchloop-mcp/identity.json.

How does Couchloop EQ handle privacy?

No emails, names, passwords, or API keys are collected. There is no analytics, no telemetry, and no third‑party data sharing. All insights are tied to anonymous session IDs.

What is the architecture behind Couchloop EQ?

It uses a modular pipeline: Request → Classify → Policy → Plan → Execute → Compose, with an intent classifier, policy engine, execution planner, tool registry (with health tracking and circuit breakers), and OpenTelemetry tracing.

「その他」の他のコンテンツ

Unity MCP ✨

justinpbarnett

Unity MCP acts as a bridge between AI assistants and your Unity Editor. Give your LLM tools to manage assets, control scenes, edit scripts, and automate tasks within Unity.

12.2K