skills.rest
IgorGanapolsky
IgorGanapolskyCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

capture-feedback

Capture structured thumbs up/down feedback with context and tags via the capture_feedback MCP tool.

25|7|Updated Mar 3, 2026
Search Tags:#agent governance#MCP tool#feedback capture#thumbs-up#thumbs-down#lesson distillation#prevent repeated failures
One-click install
npx skills add https://github.com/IgorGanapolsky/ThumbGate --skill capture-feedback
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: capture-feedback
Source: https://github.com/IgorGanapolsky/ThumbGate/tree/main/plugins/cursor-marketplace/skills/capture-feedback
Command: npx skills add https://github.com/IgorGanapolsky/ThumbGate --skill capture-feedback

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It captures thumbs up/down feedback with the context needed to reliably distill lessons from task outcomes and prevent repeated agent mistakes.

Core Features & Use Cases

  • Structured feedback capture: Records a thumbs_up or thumbs_down signal along with context, tags, and optional rubric scores.
  • History-aware lesson distillation: Uses chat history and failed tool-call context to propose what went wrong and what should change when the thumbs-down signal is vague.
  • Rule promotion pipeline support: Feeds captured feedback into a prevention-rule promotion flow so repeated failures become enforceable blocks.

Quick Start

Use the capture-feedback skill to record a thumbs_down after a failed tool call by including signal, context, tags, and up to eight prior related feedback entries if the signal is vague.

Frequently Asked Questions about capture-feedback

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I capture feedback on agent tool calls to prevent repeated mistakes?▼

To capture feedback on agent tool calls, record a structured thumbs up or thumbs down signal along with the task context and tags. This turns one-off tool call failures into reusable prevention rules to stop repeated agent mistakes.

What is the best way to distill lessons from failed tool results when the feedback is vague?▼

The best way to distill lessons from vague failed tool results is to include chat history and prior related feedback entries when recording a thumbs down signal, allowing the system to propose what went wrong and what should change.

How does recording a thumbs down signal after a test failure help with agent governance?▼

Recording a thumbs down signal after a test failure feeds the captured context into a prevention-rule promotion pipeline, transforming deployment issues into enforceable blocks for improved agent governance.

Can I include rubric scores when capturing structured feedback for completed tasks?▼

Yes, you can include optional rubric scores when capturing structured feedback for completed tasks to provide a structured quality evaluation alongside the core thumbs up or thumbs down signal and context.

When should I record feedback to turn unexpected tool results into prevention rules?▼

You should record feedback after completed tasks, unexpected tool results, test failures, deployment issues, or when explicitly requested, capturing the context needed to reliably promote repeated failures into enforceable prevention rules.

Related Skills

Explore More Software Engineering Skills→
0xmariowu0xmariowu
40

autosearch:trace-harvest

Parse session tool-call traces into structured patterns.jsonl with trace_ref provenance.

Community
Advanced
IgorGanapolskyIgorGanapolsky
25

prevention-rules

Generate and review prevention rules from failure patterns to block repeated risky AI actions.

Community
Intermediate
EmeaAppGbbEmeaAppGbb

microsoft-agent-framework

Build and orchestrate AI agents with Microsoft Agent Framework.

Official
Intermediate
rpoole-devrpoole-dev

kyc-rules

Apply KYC/AML rule grids to onboarding records and output risk ratings.

Community
Advanced