agent-feedback

Analyze Claude Code session transcripts to identify mistakes and generate corrective prompts.

Updated Feb 8, 2026

Search Tags:#compliance #CLAUDE.md #analyze #AGENTS.md #agent-debugging #session-transcripts #corrective-prompts

Ready to run

npx skills add https://github.com/Xalior/agent-skills --skill agent-feedback

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-feedback
Source: https://github.com/Xalior/agent-skills/tree/main/agent-feedback
Command: npx skills add https://github.com/Xalior/agent-skills --skill agent-feedback

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill helps users understand what happened in Claude Code agent sessions by analyzing transcripts and identifying missteps, context drift, and decision points.

Core Features & Use Cases

Session discovery: locate and summarize relevant transcripts, including metadata extraction
Corrective prompt generation: craft precise prompts to steer agents back on track
Compliance checks: verify alignment with CLAUDE.md/AGENTS.md guidance when present

Quick Start

Upload a Claude Code session transcript and request a corrective prompt.

Frequently Asked Questions about agent-feedback

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

FAQPage Schema

How do I analyze Claude Code agent session transcripts to find mistakes?▼

To analyze Claude Code session transcripts, you can upload the transcript file to extract evidence, identify decision points, and locate context drift. The skill cites relevant transcript lines to pinpoint exactly where the agent made missteps.

Can I generate corrective prompts to steer an agent back on track?▼

Yes, you can generate corrective prompts by submitting a session transcript. The skill analyzes the agent's mistakes and outputs structured prompts with clear actions and guardrails to redirect the agent effectively.

How do I check agent session compliance with CLAUDE.md and AGENTS.md guidance?▼

To check agent session compliance, provide the session transcript alongside your CLAUDE.md or AGENTS.md files. The skill verifies transcript alignment with these guidance files to detect deviations.

What is context drift in agent debugging and how do I identify it?▼

Context drift in agent debugging occurs when an agent loses sight of the original task instructions during a session. You identify it by reviewing session transcripts to locate decision points where the agent's behavior diverged from the intended path.

Does this skill require any dependencies to debug session transcripts?▼

No dependencies are required to debug session transcripts. You simply need to upload a Claude Code session transcript file to begin the analysis and corrective prompt generation process.

What is the best way to review post-session agent transcripts for errors?▼

The best way to review post-session agent transcripts is to extract cited evidence from the session file to isolate missteps and context drift. This approach clarifies what happened and generates targeted prompts for correction.