A feedback and memory layer for coding agents. Thumbs-up and thumbs-down feedback can record good patterns and mistakes for later prevention.

Use it to mark agent actions as acceptable or incorrect and build a record of patterns to avoid or follow.

It gives an agent a way to retain corrections instead of repeating the same unwanted behaviour.

thumbgate — Plugin for Claude Code and Cursor by IgorGanapolsky

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo

git clone --depth 1 https://github.com/IgorGanapolsky/ThumbGate

Made for: Claude Code, Cursor.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/plugins/igorganapolsky/thumbgate/cursor-marketplace/github.svg)](https://agentmods.dev/plugins/igorganapolsky/thumbgate/cursor-marketplace)

Your own site

<a href="https://agentmods.dev/plugins/igorganapolsky/thumbgate/cursor-marketplace"><img src="https://agentmods.dev/badge/plugins/igorganapolsky/thumbgate/cursor-marketplace/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/plugins/igorganapolsky/thumbgate/cursor-marketplace"><img src="https://agentmods.dev/badge/plugins/igorganapolsky/thumbgate/cursor-marketplace.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session not measured What this adds to a session before it is invoked.

When invoked not measured The manifest loads nothing itself; its 5 skills, 3 commands, 2 agents cost 225 tokens a session between them.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Security

Grade A, and why

thumbgate scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/cursor-marketplace/.cursor-plugin/plugin.json · 55 lines

How it starts

The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.

{
  "name": "thumbgate",
  "displayName": "ThumbGate",
  "description": "👍👎 Thumbs down a mistake — your AI agent won't repeat it. Thumbs up good work — it remembers the pattern.",
  "version": "1.36.1",
  "author": {
    "name": "Igor Ganapolsky",
    "url": "https://github.com/IgorGanapolsky"
  },
  "homepage": "https://thumbgate.ai",
  "repository": "https://github.com/IgorGanapolsky/ThumbGate",
  "license": "MIT",
  "keywords": [
    "agent-firewall",
    "ai-agent-memory",
    "coding-agent",
    "cursor",
    "cursor-plugin",
    "enterprise",
    "feedback",
    "governance",
    "grok-bot",
    "infrastructure-firewall",
    "mcp",
    "mistake-prevention",
    "pre-action",
    "pre-action-gates",
    "thumbgate",
    "thumbgate.ai",
    "thumbs-up-down"
  ],
  "logo": "assets/logo-400x400.png",
  "interface": {
    "displayName": "ThumbGate for Cursor",
    "shortDescription": "Pre-action agent firewall + reliability memory + hosted API (thumbgate.ai).",
    "longDescription": "ThumbGate is the infrastructure firewall for AI coding agents. Capture thumbs-up/down feedback, promote repeated failures into prevention rules, evaluate tool-calls before they run (local MCP + hosted API at https://thumbgate.ai), and write idempotent execution receipts for enterprise audit. Works in Cursor, Grok Bot, Claude Code, and other agent runtimes.",
    "developerName": "Igor Ganapolsky",
    "category": "Developer Tools",
    "capabilities": [
      "Interactive",
      "Write"
    ],
    "websiteURL": "https://thumbgate.ai",
    "privacyPolicyURL": "https://thumbgate.ai/privacy",
    "termsOfServiceURL": "https://github.com/IgorGanapolsky/ThumbGate/blob/main/LICENSE",
    "defaultPrompt": [
      "Check prevention gates before a deploy, publish, or destructive shell command.",
      "Capture a thumbs-down with context after a mistake so it cannot recur.",
      "Evaluate a proposed agent tool-call via ThumbGate decisions API.",
      "Search lessons for how we fixed a s

Read the full file on GitHub · 55 lines

Contents

What it installs

The manifest is a name and a version. 5 skills, 3 commands, 2 agents travel with it, and installing the plugin installs all of them — 225 tokens a session between them. Each is measured on its own page, and each can be installed alone.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

2d ago First seen · 55 lines scan A d6685898e86f

Subscribe to this mod's changes

thumbgate is a plugin published in the GitHub repository IgorGanapolsky/ThumbGate (26 stars, last pushed yesterday), licensed MIT. Its token cost is not measured: this kind of file is read by the harness, not the model. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-06.

AI Safety Building agents Guardrails Memory and context agent-reliability ai-agents developer-tools thompson-sampling

unforgit-claude-code-marketplace

Claude Code plugins for Unforgit persistent repository memory.

MiguelMedeiros/unforgit · not measured

plastic

Intent-driven state management for AI coding sessions. Neuroplasticity for your codebase.

zalom/plastic · not measured

context-management

Lifecycle of Claude Code instruction files: audit CLAUDE.md/AGENTS.md/.claude/rules instruction budget, capture session learnings, and resync docs with codebase evolution.

bengous/claude-code-plugins · not measured

titan-memory

5-layer cognitive memory for Claude Code — surprise filtering, semantic highlighting, Cortex classification, adaptive decay, cross-project learning. 30 MCP tools. 90.7% benchmark accuracy.

TC407-api/Titan-Memory · not measured

tages

Persistent codebase memory for AI coding agents. Automatically remembers conventions, architecture decisions, and project context across sessions.

ryantlee25-droid/tages · not measured

memory-loop

A memory lifecycle for Claude Code's native file-based memory: a save gate against hallucinated memories, tiered expiry (short-tier memories archived — never deleted — when they lapse), a propose-then-confirm consolidation pass (merge/dedupe long-tier memory), a periodic learning-review nudge, a two-file habit frame…

choiyounggi/groundwork · not measured