skills.rest
juanjo-zurich
juanjo-zurichCOMMUNITY
Explore all skills by this author →
GitHub

Analizador de Conversaciones

Analyzes conversation transcripts to detect problematic agent behaviors preventable with hook rules.

Updated Apr 4, 2026
Search Tags:#agent behavior#conversation analysis#severity classification#transcript review#hook rules#regex patterns
One-click install
npx skills add https://github.com/juanjo-zurich/juarvis-v4 --skill analizador-de-conversaciones-juanjo-zurich
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Analizador de Conversaciones
Source: https://github.com/juanjo-zurich/juarvis-v4/tree/main/plugins/config/skills/conversation-analyzer
Command: npx skills add https://github.com/juanjo-zurich/juarvis-v4 --skill analizador-de-conversaciones-juanjo-zurich

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI coding agents often repeat the same mistakes across sessions, and users have no systematic way to turn past frustrations into preventive rules. This Skill reviews conversation transcripts to find behaviors worth blocking with hooks before they happen again. ## Core Features & Use Cases - Frustration Signal Detection: Scans user messages for explicit corrections, frustrated reactions, and repeated complaints. - Tool-Use Pattern Extraction: Identifies which tool (Bash, Edit, Write) and which concrete command or code pattern caused each problem. - Regex Rule Generation: Converts findings into ready-to-use regex patterns with severity classification (high, medium, low) and suggested hook rules. - Use Case: After a long session where the agent repeatedly ran dangerous rm commands and added console.log to production code, run this analysis to get structured findings and suggested hook rules that prevent those behaviors in future sessions. ## Quick Start Analyze my recent conversation transcript and find behaviors that should be prevented with hook rules.

Frequently Asked Questions about Analizador de Conversaciones

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

FAQPage Schema
How do I analyze a conversation transcript for problematic agent behaviors?▼

Provide the conversation transcript and ask for a behavior analysis. The Skill reads user messages in reverse chronological order, detects corrections and frustration signals, and returns structured findings with suggested hook rules.

How to create hook rules from past AI agent mistakes?▼

The analysis converts each identified problem into a suggested hook rule containing a name, event type (bash or file), a regex pattern, and a warning message. High and medium severity findings are recommended for rule creation.

What kinds of patterns can be detected in conversation analysis?▼

It detects dangerous Bash commands like rm -rf or chmod 777, risky code patterns like eval or innerHTML in Edit/Write operations, and sensitive file paths like .env or node_modules. Each pattern is extracted as a usable regex.

Does conversation analysis flag hypothetical discussions as problems?▼

No. The Skill distinguishes edge cases such as hypothetical questions, teaching moments, and one-time accidents. These are either excluded or marked as low priority to avoid false positives.

How are detected problems categorized by severity?▼

Findings are classified as high severity (dangerous commands, security issues, data loss risks), medium severity (style violations, wrong file types), or low severity (subjective preferences). High severity items are candidates for blocking rules.

Related Skills

Explore More Software Engineering Skills→
Birdhouse-LabsBirdhouse-Labs
36

export-agent-tree

Export complete Birdhouse agent trees into a single markdown file with clickable links and transcripts.

Official
Intermediate
davila7davila7
30.5k

Writing Hookify Rules

Creates hookify rule files with YAML frontmatter and regex patterns for Claude Code hooks.

Community
Intermediate
gtappsgtapps
69

watch

Run background monitors that emit conversation notifications from subprocess outputs.

Community
Intermediate
1arley1arley

input-trust-audit

Detects server endpoints that trust client-supplied authority values instead of deriving them server-side.

Community
Intermediate