StepBack
Developers working with agentic tools
View abhijeetkumar1505/stepback-claude-plugin on GitHub- GitHub stars
- —
- Forks
- —
Project details
- Repository
- abhijeetkumar1505/stepback-claude-plugin
- Licence
- MIT
- Catalog status
- Available
- Catalog check
- 2026-11-13
- Required costs
- No Cost to use.
Repository description
Step back before you patch. StepBack is a Claude plugin for architectural reconsideration. It helps Claude distinguish a local defect from a systemic design flaw, challenge assumptions inherited from the codebase, compare alternative designs, and verify fixes against classes of edge cases. StepBack is deliberately hybrid: Skills can activate when the task context calls for architectural reconsideration. A lightweight UserPromptSubmit hook adds a reminder for bug/failure/repeated-patch prompts. It never edits files, blocks tools, or runs shell commands from project content. Slash commands provide explicit control. Specialist subagents are available in Claude Code and Cowork surfaces that support subagents. The core skill and commands are designed to be useful in Claude chat on web/Desktop too. Hooks and subagents are not available in ordinary chat.
Explore by topic
More tools in Coding & dev tools
- superpowers
obra · MIT
An open-source skills framework that gives coding agents a structured methodology (brainstorming, planning, subagent delegation) for shipping software reliably.
- ecc
affaan-m · MIT
An open-source performance layer for AI coding agents like Claude Code, Codex and Cursor, adding memory, skills and security so agents keep context and work more…
- opencode
anomalyco · MIT
OpenCode is a free, open-source AI coding agent you run from your terminal, an alternative to tools like Claude Code or Cursor that works with any LLM provider.
- ponytail
dietrichgebert · MIT
ponytail is a skill for AI coding agents that makes them behave like a lazy senior developer: write the least code needed, skip unnecessary abstraction, and stop the…