en un clic
grill-me
Relentlessly interview the user about a plan or design.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Menu
Relentlessly interview the user about a plan or design.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Basé sur la classification professionnelle SOC
Use when executing a multi-step plan and each implementation step should be delegated to subagents and independently reviewed.
Use when reviewing local changes — the working-copy diff, a branch, a commit, or a GitHub PR by number — with fresh reviewer subagents that return structured findings.
Use when a test failure, regression, exception, hang, wrong result, or unexpected behavior needs diagnosis — gathers evidence, traces the relevant path, and verifies the cause before recommending a fix.
Resolve jj (Jujutsu) conflicts. Use when jj log/status shows conflicted revisions, a rebase/squash/abandon reports 'new conflicts appeared', or files contain jj conflict markers.
Use when you want to work in an isolated jj working copy — parallel task, experimental scratch, subagent with its own tree. jj's equivalent of git worktrees: creating a workspace, working inside it from anywhere, and cleaning up without losing history.
Find things to improve in the current project
| name | grill-me |
| description | Relentlessly interview the user about a plan or design. |
Ask the user to state their plan or point to the relevant file(s) before starting.
Interview the user relentlessly about this plan until nothing remains ambiguous. Walk down each branch of the design tree, resolving dependencies between decisions one by one. Ask one question (or one tight branch of questions) per turn and wait for the answer before moving to the next.
Treat vague answers ("we'll figure it out") as unresolved — restate them concretely and ask again.
If a question can be answered by exploring the codebase, explore the codebase instead.
Provide your recommended answer only when you have enough context to have a real opinion.
When all open questions are resolved or explicitly deferred, take a fresh look at the full picture to see if there's any direction we completely missed — a concern, trade-off, or alternative that didn't come up during the branch-by-branch walk.