When is the best time to use this skill?

This skill is best used whenever progress stops, a fix fails more than twice, or when the agent feels 'lost' within a complex task or codebase.

Can it help if the context window is getting full of failed attempts?

Yes, it identifies when an agent is 'going in circles' and triggers the use of a subagent with a fresh context to handle specific subtasks.

How does it help with persistent bugs?

It employs a 'Root Cause Tracing' technique that forces the agent to document symptoms and trace logic backward to the origin rather than repeatedly attempting to fix symptoms.

What does the 'When Stuck' skill actually do?

It provides Claude with a structured framework to identify why progress has stopped and selects a specific technique—like clarification, search, or decomposition—to resume work effectively.

When Stuck

Name: When Stuck
Author: a-ariff

bya-ariff

Productivity & Workflow

Provides structured diagnostic techniques and recovery strategies to help Claude overcome development roadblocks and avoid repetitive failure loops.

About

The When Stuck skill acts as a cognitive safety net for Claude during development, preventing the common pitfall of 'looping' on failed attempts. By implementing a comprehensive dispatch table, it categorizes common development hurdles—such as unclear requirements, elusive bugs, or scope creep—and maps them to high-leverage resolution techniques like root cause tracing, semantic search, or subagent delegation. This ensures that when progress stalls, the AI stops brute-forcing and instead applies a logical, step-by-step strategy to resume productive work.

Key Features

Multi-layered search strategies for discovering logic in complex codebases
Root cause tracing workflows for persistent or 'immortal' bugs
Task decomposition patterns for managing overwhelming complexity
0 GitHub stars
Dispatch table for identifying specific types of development blockages
Anti-pattern detection to prevent brute-force attempts and context filling

Use Cases

Navigating unfamiliar repositories where the location of specific logic is unknown
Managing task complexity when a feature request begins to experience scope creep
Resolving persistent test failures that standard debugging attempts have failed to fix

About

Key Features

Multi-layered search strategies for discovering logic in complex codebases
Root cause tracing workflows for persistent or 'immortal' bugs
Task decomposition patterns for managing overwhelming complexity
0 GitHub stars
Dispatch table for identifying specific types of development blockages
Anti-pattern detection to prevent brute-force attempts and context filling

Use Cases

Navigating unfamiliar repositories where the location of specific logic is unknown
Managing task complexity when a feature request begins to experience scope creep
Resolving persistent test failures that standard debugging attempts have failed to fix