decision-interview
Skill3 starsChanged 17 days agoMITView on GitHub
Interview discipline for turning a vague intent into decisions the user can defend. One question at a time, each carrying a recommendation and its cost; look facts up yourself and put only decisions to the user; land every conclusion in the authoritative document rather than a separate log that will drift. Use whenever another skill needs to ask a user a series of design or scoping questions.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- wotjr1649/devflowtheir own version, first seen Nov 1, 2026
- microsoft/frontier-ghcp-ttt-rvastheir own version, first seen Oct 25, 2026
- Mozurok/fhorja.devtheir own version, first seen Oct 8, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- Mozurok/fhorja.dev★ 10same name
- microsoft/frontier-ghcp-ttt-rvas★ 3same name
- wotjr1649/devflow★ 1same 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 repository15 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- constraint-by-typesSkill
- decision-recordSkill
- deliverable-languageSkill
- spec-before-codeSkill
- ui-contract-assetsSkill
- ui-contract-auditSkill
- ui-contract-baselineSkill
- ui-contract-designSkill
- ui-contract-enforceSkill
- ui-contract-glossarySkill
- ui-contract-layoutSkill
- ui-contract-setupSkill
- ui-contractSkill
- ui-contract-wireframeSkill
The file
Files
Decision interview
Turn a vague intent into decisions the user can defend to someone who wasn't there. This skill is the discipline only — the questions themselves belong to whichever skill invoked this one.
Discipline
- One question at a time. Wait for the answer. Four at once makes thinking impossible, especially for non-technical users.
- Give a recommendation, and state the cost. Users need "what do I lose by picking this", not a menu. An option without a stated cost carries no information.
- Look up facts; put decisions to the user. How wide is that column, what stack is in use, where does that value live — find out yourself. Whether this should be the answer is theirs.
- When the user overrides your recommendation, accept it and list the knock-on effects. They move forward one step; you are responsible for saying which already-settled conclusions that step disturbs.
- When the user pushes back or looks confused, explain before re-asking. Never re-throw the same question verbatim.
Answering your own questions is not a liberal interpretation of this skill — it is a failure of it.
How to write the options
| Weak | Strong |
| "Option A / Option B / Option C" | Each option states: what you get, what you lose, when you'll regret it |
| "Which one do you want?" | "When it doesn't fit, which one gives way first? — giving up A costs you…, giving up B costs you…" |
| Recommendation not marked | Put the recommendation first and label it, so the user can act on the first line alone |
Costs must be specific about what will happen, not abstract. "Six months in, this becomes an infinitely-scrolling junk drawer" is a hundred times more useful than "poor extensibility".
Closing
Flatten everything into three tables and wait for confirmation before writing anything:
- Settled conclusions, numbered, one per line
- What I'll decide myself — already answered by the material at hand, or purely mechanical. Don't spend the user's attention on these
- Explicitly out of scope
Where conclusions ultimately live
That numbered table is a one-off conversational artifact. Do not turn it into another document, and do not start a separately-maintained decision log — it will drift from the authoritative document, and an agent reading two contradictory sources picks one at random.
| Content | Destination |
| "We decided to do it this way" | A rule with its reasoning, in the relevant chapter of the authoritative document |
| "We considered that and rejected it" | The authoritative document's Overturned appendix — was / became / why |
| "We're not covering that" | The authoritative document's not decided here appendix |
| "This still needs the user's call" | Ask. If you can't, record it as open and name it in conversation — don't bury it in a file |
| "Why we chose this, what we excluded, when to revisit" | A decision record. Call the Skill tool with "decision-record" |
Exception: if the user explicitly wants a shareable decision document for team alignment, write one — but put a line at the top naming which file is authoritative and stating that this one only records why, so it cannot become a second source of truth.
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.