interrogate-assumptions
Skill14 starsChanged 3 months agoMITView on GitHub
Paste this into Grok Bot after a successful run: Save the process we just used as a skill called “Interrogate Assumptions”. A useful skill states the six fields from Skills and routines. The owner (or a brief) is treating claims as facts, and those claims need to be separated into sourced, inferred, and unfounded before anyone acts on them. Every claim has a label. No unfounded item is upgraded to sourced. Disagreements are quoted, not averaged. The not-found list is…
In this repository49 other files it gives its agents.
- AGENTS.mdAGENTS.md
- convert/claude.mdCLAUDE.md
- docs/llms.txtllms.txt
- audit-subscriptionsSkill
- cited-researchSkill
- cloud-bill-noteSkill
- collate-feature-asksSkill
- coordinate-launchSkill
- daily-digestSkill
- defend-calendarSkill
- design-bot-templateSkill
- detect-docs-driftSkill
- diff-competitorsSkill
- draft-call-followupSkill
- draft-changelogSkill
- draft-issueSkill
- draft-newsletterSkill
- draft-security-answersSkill
- draft-sqlSkill
- draft-support-replySkill
- garden-memorySkill
- harden-repoSkill
- incident-timelineSkill
- install-teamSkill
- investigate-perfSkill
- log-promisesSkill
- package-evidenceSkill
- plan-tripSkill
- prep-meetingSkill
- queue-socialSkill
- ramp-new-hireSkill
- rank-account-healthSkill
- recommend-media-reallocSkill
- reconcile-expensesSkill
- remix-contentSkill
- reproduce-bugSkill
- research-accountsSkill
- review-pr-riskSkill
- route-workSkill
- sanitize-shareSkill
- scout-x-threadsSkill
- source-talentSkill
- sweep-ciSkill
- triage-inboxSkill
- vendor-renewalsSkill
- watch-aeoSkill
- watch-brandSkill
- weekly-saas-financeSkill
- write-win-lossSkill
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 23
Ask what would have to be true for the claim to be a fact.
- Distinctive: line 24
Search only the named sources plus official or primary pages in scope.
- Distinctive: line 30
Every claim has a label. No unfounded item is upgraded to sourced.
- Distinctive: line 40
Do not contact people to "confirm".
≈ “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
Interrogate Assumptions
Paste this into Grok Bot after a successful run:
Save the process we just used as a skill called “Interrogate Assumptions”.
A useful skill states the six fields from Skills and routines.
1. When to use it
The owner (or a brief) is treating claims as facts, and those claims need to be separated into sourced, inferred, and unfounded before anyone acts on them.
2. Required inputs and access
- The question, brief, memo, or list of "known" facts
- Named sources the owner believes already settle the matter
- Scope limits (dates, domains, what not to use)
3. Sequence of work
- Restate each claim as a testable sentence. Do not improve the wording into a stronger claim than the owner made.
- Ask what would have to be true for the claim to be a fact. Name the missing measurement, primary source, or date.
- Search only the named sources plus official or primary pages in scope. Quote, do not paraphrase, any number.
- Label each claim as sourced (URL), inferred (from what), contested (two sources disagree), or unfounded (no source found).
- List load-bearing assumptions the owner did not state out loud.
4. How to validate the result
Every claim has a label. No unfounded item is upgraded to sourced. Disagreements are quoted, not averaged. The not-found list is non-empty unless every claim has a primary URL.
5. What to return
Assumption map: claim, label, evidence or gap, what would falsify it, and which decisions should wait. Action log of queries.
6. What requires approval
Do not contact people to "confirm". Do not publish the map. Do not present an unsourced statistic as fact.
Type / in the composer to reference a saved skill. If it does not appear, open Settings → Plugins → Yours and enable it for this Bot.
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.