coding-audit-assumptions
Skill1 starsChanged 3 months agoMITView on GitHub
Think out of the box about a software problem or planned solution by questioning the assumptions that make it hard, then propose a different way to solve it - often simpler, with less work and less code. Use ONLY when the user explicitly invokes this skill or asks for it by name. Never trigger it automatically or as an internal step of planning, review, or remediation. Its proposals are options for the user to assess, never decisions to implement.
In this repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- ask-claudeSkill
- battery-statsSkill
- coding-consult-designSkill
- coding-explain-problemSkill
- handoffSkill
- ntd-tasksSkill
- one-by-oneSkill
- record-macos-windowSkill
- show-meSkill
- tallySkill
- video-analysisSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 20
Run this skill only when the human invoked it or explicitly asked for it in the current request.
- Distinctive: line 21
Never run it automatically, proactively, or as a step inside another workflow.
- Distinctive: line 35
Treat every proposal as speculative until the human accepts it.
- Distinctive: line 36
Do not implement a proposal, fold it into a plan, or treat it as a requirement without explicit human approval.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Audit Assumptions
Rethink the problem instead of solving it head-on. The unconventional part is the thinking, not the result: question the assumptions that make the problem (or the planned complexity) exist, and a different, better way to solve it may appear. The alternative itself can be trivially simple, boring, or just different - it only has to solve the problem better.
Main goal: hit the root, simplify
A good proposal hits the root of the problem and simplifies or outright dissolves it - so well that nothing better is possible in the current state. Usually that means less work, less code, and fewer moving parts. Do not chase cleverness; chase the shortest path to a solved problem.
Judge proposals by how completely they dissolve the problem, not by line count. Sometimes the better solution needs more code, an extra mechanism, or persistent state - that is fine when it removes the reason the problem exists. What fails this skill is complexity that only hardens the current approach against symptoms while the root cause stays.
Invocation rule (hard)
- Run this skill only when the human invoked it or explicitly asked for it in the current request.
- Never run it automatically, proactively, or as a step inside another workflow.
- "The task could benefit from it" is not an invocation.
Method
- State the problem in one sentence, as the user gave it.
- List the assumptions that make the problem or the planned complexity possible.
- For each assumption, ask: can it be removed, redesigned, or made irrelevant?
- Propose one or more alternatives that solve the stated problem better - usually with less.
- Critically assess each proposal yourself: is it genuinely adequate, or only clever-sounding? Name its weaknesses and what it gives up.
Output rule (hard)
- Present the proposals to the human for assessment, with your critical evaluation attached.
- Treat every proposal as speculative until the human accepts it.
- Do not implement a proposal, fold it into a plan, or treat it as a requirement without explicit human approval.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.