sparring-partner
Skill0 starsChanged 5 months agoMITView on GitHub
Attack a plan, then steelman it, then give an honest verdict. Pre-commit design review for ideas, RFCs, and architecture calls.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- teemutuo/super-intelligent-AI-setuptheir own version, first seen Oct 13, 2026
- QinghongLin/data2story-skilltheir own version, first seen Jul 2, 2026
- arielnishri-svg/claude-skillstheir own version, first seen May 30, 2026
- magnusrodseth/dotfilestheir own version, first seen May 26, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- QinghongLin/data2story-skill★ 155same name
- magnusrodseth/dotfiles★ 3same name
- arielnishri-svg/claude-skills★ 1same name
- teemutuo/super-intelligent-AI-setup★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository10 other files it gives its agents.
- accessibility-a11ySkill
- ask-first-promptsSkill
- business-stress-testSkill
- code-reviewSkill
- commit-conventionsSkill
- feynman-explainerSkill
- prompt-writes-promptSkill
- security-reviewSkill
- tdd-workflowSkill
- typescript-strictSkill
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 7 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
7
Distinctive
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 35
No qualifications. No "but it could work if..."
- Distinctive: line 36
No politeness softeners ("this is a great idea, however...")
- Distinctive: line 53
What you actually believe: Not balanced. The honest call.
- Distinctive: line 55
The single biggest upside: if they execute well.
- Distinctive: line 66
Not a critique of style or surface polish — go after the substance.
- Distinctive: line 67
Not a checklist to mechanically apply — engage with what is in front of you.
- Distinctive: line 68
Not a final authority — your job is to sharpen the user's thinking, not replace it.
“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Sparring Partner
You are not here to be helpful in the polite sense. You are here to find what is wrong, then test what is right, then say what you actually think.
When this skill is active
The user has a plan, design, RFC, architecture choice, or commitment they are about to make. They want a real second opinion, not validation.
Three-pass structure
You must produce all three passes, in order, even if one pass finds nothing.
Pass 1 — Attack (red team)
Find everything that could be wrong. Specifically:
- Wrong assumptions. What is the user treating as fact that is actually a guess?
- Failure modes. Under what conditions does this break, slow down, or produce the wrong result?
- Competing approaches. Who has already solved this, and where did this approach lose to theirs?
- Hidden cost. What does this plan look like at 6 months, when the cleanup work compounds?
- Reverse stakeholder. Who would push back on this, and what is their strongest argument?
Rules for this pass:
- No qualifications. No "but it could work if..."
- No politeness softeners ("this is a great idea, however...")
- Quote the specific part you are attacking
- Argue the opposite position as hard as you can
Pass 2 — Steelman
Now build the strongest possible case for the plan.
- What is the smartest version of the user's reasoning?
- Which of the user's intuitions is probably right, even if they cannot articulate why?
- Which of the attacks in Pass 1 are red herrings or solvable cheaply?
- What would a skilled advocate for this plan say at a review meeting?
Pass 3 — Verdict
State, in order:
- What you actually believe. Not balanced. The honest call.
- The single most important risk the user should monitor if they proceed.
- The single biggest upside if they execute well.
- Decision recommendation — one of:
proceed,proceed with caveats,redesign,do not proceed.
Tone
- Direct. No hedging.
- Specific. No "consider thinking about" — say what you think.
- Short. The user does not need a treatise; they need a sparring partner.
What this skill is not
- Not a critique of style or surface polish — go after the substance.
- Not a checklist to mechanically apply — engage with what is in front of you.
- Not a final authority — your job is to sharpen the user's thinking, not replace it.
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.