grill-me
Skill1 starsChanged 4 months agoMITView on GitHub
Knowledge-extraction interview. Use when Alex says "grill me on X", "extract what I know about X", "get this out of my head", or before writing any strategy doc, brief, or positioning where Alex's tacit knowledge is the real input. 15-30 adaptive questions, ONE at a time, then a structured brief to knowledge/brainstorms/ plus routed updates to entities/tacit files. EXTRACTS knowledge FROM Alex — NOT a quiz that tests Alex, and NOT /roast (which attacks a plan).
193 other projects use a version of this.
Adopted by
Most recently taken up by
- bamlab/flashlighttheir own version, first seen Nov 16, 2026
- philoserf/claude-code-configtheir own version, first seen Nov 15, 2026
- niki914/zafirotheir own version, first seen Nov 15, 2026
- quickchatai/quickreviewtheir own version, first seen Nov 15, 2026
- HiteSit/real-scientist-skillstheir own version, first seen Nov 15, 2026
- alexj11324/Mikkatheir own version, first seen Nov 15, 2026
Used with
Projects with it also use brief-to-tasks, writing-beats and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
julianoczkowskiSkillbrief-to-tasks
Break a design brief into an ordered checklist of independently buildable tasks using vertical slices. Saves as a markdown checklist. Use when user wants to break down work, create tasks from a brief, plan implementation order, or mentions "tasks" or "breakdown".
Also in 7 projects that use this
LesaniSkillwriting-beats
Shape an article as a journey of beats, choose-your-own-adventure style. The user picks a starting beat from the raw material, you write only that beat, then offer options for where to pivot next, beat by beat, until the article reaches a natural end. Use when the user has raw material and wants to assemble it as a narrative rather than an argument.
Also in 74 projects that use this
simo97Skillmigrate-to-shoehorn
Migrate test files from `as` type assertions to @total-typescript/shoehorn. Use when user mentions shoehorn, wants to replace `as` in tests, or needs partial test data.
Also in 78 projects that use this
simo97Skillrequest-refactor-plan
Use this skill when user wants to create a refactor plan.
Also in 45 projects that use this
LesaniSkillwriting-shape
Take a markdown file of raw material and shape it into an article through a conversational session — drafting candidate openings, growing the piece paragraph by paragraph, arguing about format (lists, tables, callouts, quotes) at each step. Use when the user has a pile of notes, fragments, or a rough draft and wants help turning it into something publishable.
Also in 73 projects that use this
RoentekSkillloop-me
Grill me about specs for the workflows I want to build, within this workspace.
Also in 64 projects that use this
- +1 more from simo97
- +1 more from Lesani
Other versions193 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 193
- mattpocock/skills★ 325ksame name
- The-Vibe-Company/quivr★ 40ksame name
- QuivrHQ/quivr★ 40ksame name
- langfuse/langfuse★ 37ksame name
- nrwl/nx★ 29ksame name
- windmill-labs/windmill★ 19ksame name
- GreyDGL/PentestGPT★ 16ksame name
- humanlayer/humanlayer★ 13ksame name
- jaredpalmer/kev★ 9.8ksame name
- openstatusHQ/openstatus★ 9.3ksame name
183 more not listed
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 repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- commsSkill
- composeSkill
- daily-briefSkill
- domain-hygieneSkill
- idea-deep-diveSkill
- laila-headless-conductSkill
- laila-judgmentSkill
- laila-ops-debuggingSkill
- level-upSkill
- new-domainSkill
- os-auditSkill
- personaSkill
- quarterly-refreshSkill
- roastSkill
- running-briefSkill
- session-wrapSkill
- setupSkill
- specSkill
- validate-ideaSkill
- whats-nextSkill
The file
We split this skill into the rules it gives an agent
12 rules. 0 are common, 10 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 179 edited versions we read keep these.
10
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
BEFORE writing any strategy doc, positioning piece, or client brief where Alex's tacit knowledge is the actual input — run this first, …
- Very specific: line 18
NOT for: testing Alex's recall (that's a quiz — this skill has no answer key), critiquing a plan (/roast), or finding automation candidates …
- Distinctive: line 47
Never answer your own question or think aloud toward an answer.
- Distinctive: line 48
Never embed a candidate answer in the question ("Was it because of budget?" → produces the agent's knowledge, not Alex's).
- Distinctive: line 49
Never batch. One question, one answer, one follow-up decision.
- Distinctive: line 50
When Alex contradicts an earlier answer, surface it neutrally and let them resolve it — a live contradiction is high-value material, not an …
- Very specific: line 56
Write knowledge/brainstorms/<topic-slug>-YYYY-MM-DD.md with these sections:
- Distinctive: line 94
Before declaring this skill done (and again after edits):
- Distinctive: line 95
One-at-a-time test: run a session; any message containing two questions = fail.
- Distinctive: line 98
Traceability test: every claim in "What Alex knows" cites a question number.
- Distinctive: line 103
After each run: if Alex corrected the flow (questions too long, follow-ups missing an obvious vague noun, brief mis-organized), this skill …
- Distinctive: line 104
If the same correction happens twice, the fix is MANDATORY before the next run.
“Common” is measured: kept by at least 6 in 10 of the 179 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
Grill Me
Alex knows things that exist nowhere in the system — client dynamics, market instincts, hard-won lessons from failed attempts. Any strategy doc written without that knowledge is written from the agent's priors, not Alex's experience. This skill gets it out of Alex's head and into the knowledge layer, one relentless question at a time.
Setup note: the output directory
knowledge/brainstorms/may not exist yet —mkdir -p knowledge/brainstormson first use.
When to use
- "grill me on the Acme phase-2 expansion", "extract what I know about pricing consulting work", "get this out of my head"
- BEFORE writing any strategy doc, positioning piece, or client brief where Alex's tacit knowledge is the actual input — run this first, write second
- NOT for: testing Alex's recall (that's a quiz — this skill has no answer key), critiquing a plan (
/roast), or finding automation candidates (/level-up)
Steps
Phase 1 — Scope
One topic per session. If $ARGUMENTS names it, confirm in one line and start. If not, ask for it. Reject scope creep mid-interview: "That's a second topic — parking it for another session" (and note it in the brief's open threads).
Before the first question, silently check what the system already holds: relevant knowledge/entities/ files, knowledge/tacit/, the domain's tracking/status.md. The gap between what's on file and what Alex knows IS the interview target. Never ask for facts a file already answers — that wastes budget and teaches Alex the interview is theater.
Phase 2 — The interview
15-30 adaptive questions. ONE question per message, never a batch. Each question under ~2 sentences. Budget check at ~15: if answers are still producing new material, keep going; hard stop at 30.
Follow-up triggers — these are the skill:
| Signal in Alex's answer | Your next question |
| Vague noun ("the usual stakeholder problems", "some friction") | Name it: "Which stakeholders, and what did they actually do?" |
| An opinion ("fixed-fee is a trap") | The incident behind it: "What happened that taught you that?" |
| "Obviously" / "everyone knows" / "as usual" | Dig HERE — obvious-to-Alex is exactly what's missing from the system: "Obvious how? Walk me through it like I'm new." |
| A generalization ("clients always...") | A specific case: "Most recent time that happened — who, when?" |
| A number-shaped claim without a number ("it took forever", "a big chunk of revenue") | The number: "How long? What percentage?" |
| A success story | The failed attempts before it: "What did you try first that didn't work?" |
Chase specifics relentlessly: names, numbers, dates, sequences, what was tried and abandoned. A claim without an incident behind it is an instinct; note it as one, then probe for the incident anyway.
Interview discipline (hard rules):
- Never answer your own question or think aloud toward an answer. Silence after the question; wait.
- Never embed a candidate answer in the question ("Was it because of budget?" → produces the agent's knowledge, not Alex's). Ask open: "Why do you think that happened?"
- Never batch. One question, one answer, one follow-up decision.
- When Alex contradicts an earlier answer, surface it neutrally and let them resolve it — a live contradiction is high-value material, not an error to smooth over.
Stop conditions: answers start repeating, Alex says stop, or the 30-question budget is spent. Stopping early on repetition is a win — it means the vein is mined.
Phase 3 — Synthesis
Write knowledge/brainstorms/<topic-slug>-YYYY-MM-DD.md with these sections:
- The question that prompted this — why the extraction happened, what it feeds (e.g., "Acme phase-2 expansion strategy doc, due next week").
- What Alex knows — organized claims, grouped by theme, each traceable to a specific answer ("Q7"). Distinguish incident-backed claims from instincts Alex couldn't source.
- The incidents — the concrete stories behind the claims, with names/numbers/dates as given. These are the evidence layer; keep them verbatim-close.
- Open contradictions and uncertainty — anything Alex flagged as unsure, plus contradictions surfaced in Phase 2 and how (or whether) Alex resolved them. Parked second topics land here too.
- What the system should do with this — a routing table: which fact goes to which
knowledge/entities/file, which lesson to whichknowledge/tacit/file, which status change to whichtracking/status.md. Proposed, not auto-applied.
Phase 4 — Route durable facts
Walk the routing table with Alex:
- Tier 1 (auto-execute + notify): updates to Alex-only knowledge files —
knowledge/entities/,knowledge/tacit/. Apply, then notify via the command channel perconfig/autonomy-rules.json. - Tier 3 (propose-and-wait): anything beyond the knowledge layer — status docs feeding shared views, CRM notes, anything visible to others. Present the diff, wait.
Close with a one-screen summary in chat: topic, question count, top 3 extracted claims, where the brief lives, what was routed where.
Worked exchange (the drill-down pattern)
Laila: What made the Acme phase-2 expansion harder to scope than the original engagement? Alex: Honestly the usual stakeholder problems — procurement got involved late. Laila: "Usual stakeholder problems" — which stakeholders, and what did each actually do? Alex: Sam Okafor champions us internally but doesn't own the decision. Jordan owns the engagement commitment and obviously cares about fee-to-scope fit. Laila: You said "obviously" — obviously how? What has Jordan said or done that tells you the engagement fee is the lever? Alex: Last March they dropped a vendor mid-engagement when the fee ran 30% past the scoped work. Sam told me over lunch.
Three turns took "the usual problems" to a dated incident with names and a number — which is now routable to
knowledge/entities/jordan-lee.md.
Rules
- One topic per session. One question per message. Questions under ~2 sentences.
- Extraction, not examination: there is no grading, no score, no "correct" answer. Alex is the source of truth about what Alex knows.
- Leading questions are contamination — if you catch yourself supplying a hypothesis inside a question, re-ask it open.
- Writes: the brainstorm brief (always), Tier 1 knowledge-file updates (with notify), nothing else without approval.
- The brief must survive Alex forgetting everything — six months later, the incidents section alone should reconstruct why each claim was believed.
Verification (MANDATORY)
Before declaring this skill done (and again after edits):
- One-at-a-time test: run a session; any message containing two questions = fail.
- Drill-down test: answer a question with "the usual client stuff" — the next question must name and chase the vague noun, not move on.
- No-leading test: rereading the session, no question may contain a candidate answer or hypothesis.
- Traceability test: every claim in "What Alex knows" cites a question number; every incident has at least one specific (name, number, or date) or is explicitly marked as an unsourced instinct.
- Routing-gate test: nothing outside
knowledge/brainstorms/,knowledge/entities/,knowledge/tacit/is written without an explicit Tier 3 approval.
Self-improvement (MANDATORY)
- After each run: if Alex corrected the flow (questions too long, follow-ups missing an obvious vague noun, brief mis-organized), this skill is wrong, not Alex. Propose the edit immediately, or log
{"skill": "grill-me", "correction": "...", "date": "YYYY-MM-DD"}tostate/skill-feedback.jsonl. - If the same correction happens twice, the fix is MANDATORY before the next run.
- Track which follow-up triggers actually produced incidents; if a new signal pattern recurs across sessions, add it to the trigger table.
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.

