Use this when debugging feels stuck, circular, or is giving misleading signals — especially when the problem crosses boundaries like permissions, sandboxing, IPC, app extensions, files, networking, async lifecycle, OS APIs, or multiple processes. Instead of repeatedly tweaking the same broken path, keep the end goal fixed and swap the mechanism for something simpler or fundamentally different, testing one assumption at a time to reveal which layer is actually failing. Trigger this whenever you notice yourself making the same fix repeatedly, getting inconsistent results from the same approach, or suspecting the bug is one layer deeper than where you're currently poking at it.
What specific approach are you using to achieve the goal?
Ask: "If I couldn't use this mechanism, how else could I accomplish the same goal?"
Brainstorm 2–5 fundamentally different pivots.
Prefer pivots that remove complexity or test one assumption.
They do not have to be production-ready.
Choose the cheapest pivot.
The goal is to learn something quickly, not build the final solution.
Execute the pivot.
Observe what changes.
Record what assumptions were confirmed or eliminated.
Either:
Keep the pivot if it's better.
Or use what you learned to improve the original implementation.
Repeat until the goal is achieved.
Principles
Preserve the goal, not the implementation.
Every pivot should teach you something.
Prefer experiments over theories.
Reduce assumptions whenever possible.
Temporary solutions are valuable if they increase understanding.
1---2name: goal-preserving-pivoting3description: Use this when debugging feels stuck, circular, or is giving misleading signals — especially when the problem crosses boundaries like permissions, sandboxing, IPC, app extensions, files, networking, async lifecycle, OS APIs, or multiple processes. Instead of repeatedly tweaking the same broken path, keep the end goal fixed and swap the mechanism for something simpler or fundamentally different, testing one assumption at a time to reveal which layer is actually failing. Trigger this whenever you notice yourself making the same fix repeatedly, getting inconsistent results from the same approach, or suspecting the bug is one layer deeper than where you're currently poking at it.4---561. Define the real goal.7 - Ignore the current implementation.8 - State the desired outcome in one sentence.92. Identify the current mechanism.10 - What specific approach are you using to achieve the goal?113. Ask: "If I couldn't use this mechanism, how else could I accomplish the same goal?"124. Brainstorm 2–5 fundamentally different pivots.13 - Prefer pivots that remove complexity or test one assumption.14 - They do not have to be production-ready.155. Choose the cheapest pivot.16 - The goal is to learn something quickly, not build the final solution.176. Execute the pivot.18 - Observe what changes.19 - Record what assumptions were confirmed or eliminated.207. Either:21 - Keep the pivot if it's better.22 - Or use what you learned to improve the original implementation.238. Repeat until the goal is achieved.2425## Principles26- Preserve the goal, not the implementation.27- Every pivot should teach you something.28- Prefer experiments over theories.29- Reduce assumptions whenever possible.30- Temporary solutions are valuable if they increase understanding.
Run npx skillmds@latest add aryanrogye/goal-preserving-pivoting in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
Use this when debugging feels stuck, circular, or is giving misleading signals — especially when the problem crosses boundaries like permissions, sandboxing, IPC, app extensions, files, networking, async lifecycle, OS APIs, or multiple processes. Instead of repeatedly tweaking the same broken path, keep the end goal fixed and swap the mechanism for something simpler or fundamentally different, testing one assumption at a time to reveal which layer is actually failing. Trigger this whenever you notice yourself making the same fix repeatedly, getting inconsistent results from the same approach, or suspecting the bug is one layer deeper than where you're currently poking at it. It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
AryanRogye (@aryanrogye) published this skill. Their other Agent Skills are listed on their SkillMD profile.
AryanRogye