From thumbgate
Provides pre-action gates for AI coding agents: captures thumbs-up/down feedback, auto-generates prevention rules from repeated failures, blocks risky tool calls via MCP PreToolUse hooks. Trigger for safety guardrails, structured feedback, or DPO exports.
npx claudepluginhub igorganapolsky/thumbgateThis skill uses the workspace's default tool permissions.
Pre-action gates that stop AI coding agents from repeating known mistakes.
Recalls past mistakes and prevention rules before tasks, captures success/failure feedback after, to prevent agents repeating errors. Uses local MCP tools like recall and capture_feedback.
Enforces tiered runtime guardrails on Claude Code agent actions: auto-approves reads/routines, notifies on writes/commits, requires approval for installs/emails/deletes, blocks credential leaks/sudo. Logs audits.
Provides design patterns for autonomous coding agents: agent loops, tool calling APIs, permissions, browser automation, human-in-the-loop workflows. Python examples included.
Share bugs, ideas, or general feedback.
Pre-action gates that stop AI coding agents from repeating known mistakes.
npx thumbgate init
This installs the MCP server and wires it into your agent's tool configuration. No API keys required for the free tier.
Or install globally:
npm install -g thumbgate
thumbgate init
Add to your agent's MCP config (e.g., claude_desktop_config.json or .cursor/mcp.json):
{
"mcpServers": {
"thumbgate": {
"command": "npx",
"args": ["-y", "thumbgate"]
}
}
}
When an agent action succeeds or fails, capture structured feedback:
After a failure pattern repeats (default: 2 occurrences), ThumbGate auto-generates a prevention rule. These rules are injected into agent context before every tool call, blocking the known-bad pattern before it executes.
Gates intercept tool calls via the MCP PreToolUse hook. Each gate checks the proposed action against:
If a match is found, the gate blocks execution and surfaces the prior failure context.
Bounded retrieval of relevant feedback history for the current task. The agent gets exactly the lessons that matter, not the entire history.
| Tool | Description |
|---|---|
capture_feedback | Record thumbs-up/down with structured context |
search_lessons | Query the lesson DB by keyword or semantic similarity |
get_prevention_rules | Retrieve active prevention rules for the current context |
session_primer | Get session handoff context (last task, next step, blockers) |
export_dpo | Export feedback pairs in DPO training format |
| Free | Pro | Team | |
|---|---|---|---|
| Feedback capture | 3/day | Unlimited | Unlimited |
| Lesson search | 5/day | Unlimited | Unlimited |
| Active gates | 5 | Unlimited | Unlimited |
| Dashboard | - | Yes | Yes |
| DPO export | - | Yes | Yes |
| Seats | 1 | 1 | Per-seat |
| Price | $0 | $19/mo | $49/seat/mo |
Start a 7-day free trial of Pro: https://thumbgate-production.up.railway.app/go/pro?utm_source=skill
ThumbGate works with any MCP-compatible agent: