agent-llm-stability

Community

Stability diagnostics for agent LLMs

Authorkrikz
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Specialized tooling to diagnose and fix degradation, hangs, and context drift in an autonomous agent's LLM-driven cycle (dialogue_node.py + MCP tools). It helps prevent cascading failures when the agent becomes unresponsive or memory or context grows unmanageably, ensuring reliable operation in production.

Core Features & Use Cases

  • Diagnostics for deadlocks, memory leaks, prompt issues, and tool-call reliability across the agent loop (STT, dialogue_node, MCP tools, and conversation_history).
  • Guided workflows to reproduce, isolate, and fix order-independent failures such as timeouts, slow streams, and unintended memory accumulation, with safe rollbacks.
  • Use Case: When the agent stalls after extended dialogues or drifts into unrelated topics, this skill provides targeted fixes to restore stability.

Quick Start

Run an initial diagnostic pass on the agent thread to detect where the instability originates, then apply recommended fixes in sequence.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: agent-llm-stability
Download link: https://github.com/krikz/rob_box_project/archive/main.zip#agent-llm-stability

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.