skills.rest
ybbms777
ybbms777COMMUNITY
Explore all skills by this author →
GitHub

self-eval

Detect user corrections, tool failures, and pauses in conversational sessions.

4|Updated Apr 3, 2026
Search Tags:#embeddings#audit#reflection#lancedb#session-end#self-evaluation
One-click install
npx skills add https://github.com/ybbms777/openclaw-claude-code-integration --skill self-eval-ybbms777
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: self-eval
Source: https://github.com/ybbms777/openclaw-claude-code-integration/tree/main/skills/self-eval
Command: npx skills add https://github.com/ybbms777/openclaw-claude-code-integration --skill self-eval-ybbms777

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires lancedb, and includes scripts (resource) components.

What problem does it solve?

This skill automatically performs a post-session self-evaluation to detect user corrections, tool failures, and rule-triggered pauses, and writes reflections to LanceDB for learning and auditing. If no anomalies are found, it exits silently.

Core Features & Use Cases

  • Detect explicit and implicit user corrections from recent session messages.
  • Capture tool invocation failures and BDX-related errors, plus pause confirmations, and convert them into structured reflections.
  • Persist learnings from corrections to LEARNINGS.md and store validated reflections with embeddings for efficient retrieval.

Quick Start

End a session to automatically trigger the self-eval hook, or run the script with --dry to test its behavior.

Frequently Asked Questions about self-eval

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

FAQPage Schema
How do I automate end-of-session self-evaluation for conversational agents?▼

Automate end-of-session self-evaluation by detecting user corrections, tool failures, and rule-triggered pauses, then writing structured reflections to LanceDB for auditing. If no anomalies are found, it exits silently.

Can I store conversational reflections with embeddings in LanceDB?▼

Yes, you can store validated reflections with embeddings in LanceDB using SiliconFlow or a minimax fallback for efficient retrieval and auditing.

What is the best way to detect user corrections from session messages?▼

Detect explicit and implicit user corrections from recent session messages by running an automated self-evaluation script at session-end.

How do I test self-evaluation behavior without ending a session?▼

Test self-evaluation behavior without ending a session by running the script with the --dry flag to simulate its detection and logging logic.

Does the self-evaluation script export tool failures and corrections to a log?▼

Yes, the self-evaluation script exports detected tool failures and user corrections to LEARNINGS.md for persistent logging and audit.

Related Skills

Explore More Data & Analytics Skills→
cwinvestmentscwinvestments
412

echo

Retrieve historical AI session data via semantic vector and keyword search.

Community
Advanced
ybbms777ybbms777
4

evolve

Extract recurring patterns from LanceDB reflections and LEARNINGS.md into MUST/NEVER/ALWAYS governance rules.

Community
Advanced
luigiinredluigiinred
1

always-use-askquestion

Enforce AskQuestion tool usage for all structured user interactions.

Community
Basic
ITSpecialist111ITSpecialist111
43

rfp-answer-bank

Manage RFP Q&A entries with versioning, deprecation, and audit logging.

Community
Advanced