What happens if the third attempt also fails?

If the third attempt fails after research, the skill provides an escalation template to help the AI clearly articulate the blocker to a human developer.

When is the Research After Failure skill triggered?

It is automatically triggered after two consecutive failed attempts at solving a specific problem, such as failing tests or build breaks.

How does it prevent the AI from researching indefinitely?

The skill includes a built-in time-boxing protocol, limiting research to a maximum of 30 minutes total (divided between local and online search) before requiring action or escalation.

Does it communicate findings to the team?

Yes, it is designed to synthesize research findings and post them as a comment on the relevant GitHub issue, including the proposed new approach.

What sources does this skill research?

The skill systematically checks READMEs, local docs, the repository's git history, existing code patterns, and online resources like Stack Overflow and official library documentation.

Research After Failure

Name: Research After Failure
Author: troykelly

bytroykelly

생산성 및 워크플로

Enforces a structured research protocol after two consecutive failed attempts to prevent repetitive errors and ensure informed problem-solving.

소개

Research After Failure is a strategic workflow enhancement for Claude Code designed to break loops of unproductive trial-and-error. When the agent fails to solve a problem twice in a row, this skill triggers a mandatory 'stop-and-research' phase, requiring the AI to document past failures, analyze repository documentation, search the existing codebase for patterns, and consult online resources before attempting a third solution. This systematic approach ensures that subsequent attempts are based on hard data rather than guesswork, significantly improving success rates for complex debugging tasks while providing clear escalation paths to human developers when blockers persist.

주요 기능

Mandatory documentation of failed approaches and error patterns
Automated GitHub issue updates with research findings and new action plans
0 GitHub stars
Comprehensive search across local documentation, codebase, and web resources
Strict time-boxing for research phases to maintain development momentum
Structured research protocol triggered after two consecutive failures

사용 사례

Breaking loops of repetitive failing test cases during complex refactors
Discovering established implementation patterns within large, unfamiliar codebases
Solving obscure bugs that require deep diving into external library documentation

소개

주요 기능

Mandatory documentation of failed approaches and error patterns
Automated GitHub issue updates with research findings and new action plans
0 GitHub stars
Comprehensive search across local documentation, codebase, and web resources
Strict time-boxing for research phases to maintain development momentum
Structured research protocol triggered after two consecutive failures

사용 사례

Breaking loops of repetitive failing test cases during complex refactors
Discovering established implementation patterns within large, unfamiliar codebases
Solving obscure bugs that require deep diving into external library documentation