Agent Performance Reviewer is a free, MIT-licensed agent skill that reads your AIML.chat conversation logs and surfaces patterns — what the agent answers well, where it fails, where it escalates, and what to fix next. It turns raw chat history into a clear improvement plan.
What it does
The skill analyzes a batch of conversations and clusters them: the questions the agent answered confidently with citations, the ones it couldn't and had to escalate, the topics where visitors pushed back, and the moments it nearly hallucinated but caught itself. It quantifies deflection, escalation, and lead-capture rates, and ranks the failure topics by frequency so you know exactly what to document, re-prompt, or ground better. Rather than reading transcripts one by one, you get a structured review with concrete actions. Run it periodically to track whether your changes actually moved the numbers.
The skill
name: agent-performance-reviewer
Use to find patterns in agent conversation logs.
Steps:
1. Pull a batch of conversation logs from AIML.chat analytics.
2. Cluster into answered, escalated, pushed-back, and risky.
3. Quantify deflection, escalation, and lead-capture rates.
4. Output ranked failure topics and concrete fixes.
How to use it
Copy the SKILL.md into your agent framework's skills folder and run it on exported logs, or review the patterns in AIML.chat's analytics. It pairs with the widget analytics reporter for a shareable summary and the knowledge base auditor to close content gaps. The agent overview explains the behaviors it measures.
Ready to improve your agent? Start free, review your logs, and ship fixes in minutes — see pricing for analytics limits.
What patterns does it find?+
What the agent answers well, where it escalates, where visitors push back, and the failure topics ranked by frequency.
Can I track improvement over time?+
Yes — run it periodically to see whether your prompt and content changes actually moved deflection and escalation rates.
Is the skill free?+
Yes — it is MIT-licensed and free to copy. Analytics depth depends on your plan; see pricing.