method-grill skill
✓ VerifiedInterviews researchers one question at a time to stress-test study goals, hypotheses, data, and analysis plans before submission.
- GitHub stars
- 6
- Files
- 5
- Last commit
- Oct 31, 2026
- License
- MIT
- Allowed tools
- —
What does the method-grill skill do?
method-grill interviews researchers one question at a time to stress-test study goals, hypotheses, data, and analysis plans before submission.
Summary generated from the skill's SKILL.md.
About the method-grill skill
Researchers use this skill to stress-test study goals, hypotheses, data, and analysis plans through a structured interview process before submitting a Statistical Analysis Plan or manuscript. By treating a research project as a decision tree with seven branches, the tool identifies potential weaknesses and reviewer objections in areas like population definitions, measurement validity, and journal scope fit. It functions as a rigorous conversational partner that asks one decision-changing question at a time, providing a recommended answer and rationale for each to help guide the user toward a more defensible study design.
An agent should load this skill when a user says "grill me" or needs to validate a research idea. The tool relies on existing project files like research-question.yaml or SAP drafts and uses PubMed and Paperclip MCP tools to look up literature facts, event rates, and journal guidelines. It avoids asking for information it can find independently. The process produces a standalone markdown file in docs/brainstorms/ containing a summary of decisions and identified risks, or it returns confirmed decisions and unresolved risks to the method-plan skill during integrated planning workflows.
What an agent can do with it
- 1Stress-test a research hypothesisAsk the agent to challenge a study's hypothesis by evaluating the population, exposure, comparator, and expected direction.
- 2Identify potential reviewer objectionsRequest a grill session to uncover specific weaknesses in an analysis plan that a hostile reviewer might attack.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the method-grill skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the method-grill skill in Plugin
/plugin marketplace add sajor2000/methodsmith /plugin install methodsmith@methodsmith
Install the method-grill skill in Claude Code
~/.claude/skills/method-grill/SKILL.md
$ npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill ~/.claude/skills/method-grillInstall the method-grill skill in Claude Code (project)
.claude/skills/method-grill/SKILL.md
$ npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill .claude/skills/method-grillInstall the method-grill skill in Codex
~/.agents/skills/method-grill/SKILL.md
$ npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill ~/.agents/skills/method-grillInstall the method-grill skill in Gemini CLI
~/.gemini/skills/method-grill/SKILL.md
$ npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill ~/.gemini/skills/method-grillInstall the method-grill skill in Claude.ai
npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill method-grill && zip -r method-grill.zip method-grill # claude.ai → Customize → Skills → + → Create skill → Upload a skill → method-grill.zip
Automated checks
Ran Nov 15, 2026Full report →Passed our automated checks. That is not a security audit: read the scripts before you let an agent run them. How checks work
- Source repository linked and reachablesajor2000/methodsmith/plugins/methodsmith/skills/method-grill is public and reachablegithub.com/sajor2000/methodsmith/tree/HEAD/plugins/methodsmi ↗Pass
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit 2026-10-31github.com/sajor2000/methodsmith/commits ↗Pass
- No download-and-run or credential theft in scriptsNo scripts in the skill folderPass
- No hidden instructions in SKILL.mdNo override phrases, safeguard switches, credential requests or hidden charactersgithub.com/sajor2000/methodsmith/blob/HEAD/plugins/methodsmi ↗Pass
- Pre-approved tools are scopedNo pre-approved tools; the agent asks before using onegithub.com/sajor2000/methodsmith/blob/HEAD/plugins/methodsmi ↗Pass
- SKILL.md follows the Agent Skills formatValid name and description; says when to use it; 80 linesgithub.com/sajor2000/methodsmith/blob/HEAD/plugins/methodsmi ↗Pass
SKILL.md
View on GitHubGrill a Research Project
Skill Value
- Problem it solves: Studies fail at review because the goal, hypothesis, data, analysis, and journal fit were never challenged together before the SAP was written.
- Use when: The user says grill me, or wants a study idea, hypothesis, or analysis plan stress-tested before planning or submission.
- Output: Standalone:
docs/brainstorms/<YYYY-MM-DD>-<slug>-grill.mdwith decisions and risks. In planning integration: confirmed decisions and unresolved risks returned tomethod-plan. - Ask only if: The answer is a decision only the researcher can make; look up every fact yourself first.
- Do not do: Do not write the SAP, run analyses, open patient-level data, or act on the decisions before the user confirms a shared understanding.
- Interaction: Check repo/config/chat evidence first. Ask one decision-changing question at a time; use the current harness's blocking question UI when available, otherwise present numbered choices and wait.
How the grill works
Treat the project as a decision tree with seven branches. Every decision unlocks the decisions that depend on it; the frontier is the set of open decisions whose prerequisites are already settled.
| Branch | The decision it forces |
|---|---|
| Goal | What decision, practice, or policy changes if the study succeeds, and for whom |
| Task and deliverable | What to produce, for whom, under which constraints, and how success will be judged |
| Hypothesis | The falsifiable question: population, exposure or intervention, comparator, outcome, time zero, and expected direction |
| Data | Whether this source can answer it: provenance, grain, access and approvals, coverage, missingness, and measurement validity |
| Proposed analysis | Design, estimand, confounding or causal assumptions, missing-data plan, multiplicity, sensitivity analyses, and sample size |
| Target journal | Scope fit, article type, reporting checklist, word and display limits, and data-sharing policy |
| Audience | Who must believe the result, what they already believe, and what a hostile reviewer will attack first |
Read references/question-bank.md for the probes, strong-answer criteria, and red flags for each branch.
Evidence access
Before any literature lookup, read references/evidence-access.md (PubMed and Paperclip MCP tools, fallbacks, and evidence rules); never invent a citation, effect size, or guideline item.
- Look up here: the facts the interview must never ask the user for: prior studies on the hypothesis, published event rates, validated exposure and outcome definitions (with their PPV or sensitivity), and the target journal's recent similar papers (search with the journal name). Journal author guidelines (scope, article types, limits, required checklists) come from web search, not PubMed.
Workflow
Planning integration
When method-plan invokes this skill, use the decision tree and question bank for decisions that affect the requested plan. First establish the user's goal, task and deliverable, question or hypothesis when applicable, data, and proposed methods from existing evidence. Check settled choices for consequential weaknesses, not just missing fields; explain the specific concern before asking a decision-changing question. Do not re-ask settled choices without a reason. For descriptive or exploratory work, clarify the aim without requiring a directional hypothesis. Journal and audience questions are conditional on their relevance to the task.
Resolve the sensitive-data boundary as in standalone step 2 before data access; use schema-only when unset in non-interactive mode. Look up accessible facts within that boundary; ask the researcher only for decisions or facts unavailable to you. Ask one question at a time with a recommended answer and rationale, and recompute dependent questions after each answer. Stop when the planning-critical decisions are clear or the user stops. Summarize choices and risks, ask the user to confirm the shared understanding, then return confirmed choices and unresolved risks to method-plan. If the user stops, carry unconfirmed choices as gaps. Do not create a separate grill record or route back to method-plan as a new invocation. The user may request a standalone grill record separately. In non-interactive mode return unresolved choices as gaps without inventing answers.
When comparing methods, state how each changes the population, time zero, estimand, and assumptions where relevant. Do not treat a landmark design as interchangeable with a baseline target-trial strategy or promise that a method eliminates bias. Aligning eligibility and time zero alone does not resolve treatment assignment over a grace period. Until the drug, comparator, treatment window, and relevant data are known, recommend clarifying those decisions rather than selecting a specific causal estimator. If the facts do not support a responsible recommendation, explain the uncertainty and ask for the information needed rather than inventing one.
Preserve the outcome as stated: do not add all-cause, in-hospital, or other qualifiers unless supported or confirmed. An active comparator can improve comparability when clinically appropriate, but does not inherently reduce confounding; explain the assumptions and remaining confounding rather than promising an advantage.
Standalone interview
-
Gather facts before asking. Read whatever exists:
analysis/research-question.yaml,docs/brainstorms/, any SAP (analysis/sap.mdor a file withsap_versionfrontmatter),reports/data-qa/oranalysis/data-qa-report.md, and.methodsmith/config.local.yaml. Look up journal scope, article types, word limits, and required checklists from the journal's author guidelines with available search tools instead of asking. Look up literature facts (prior studies, event rates, validated definitions, the journal's recent similar papers) with the Evidence access tools above, reading each source before citing it. Use parallel subagents for independent lookups when the platform supports them. Never ask the user for something you can find. -
Sensitive-data boundary. Resolve
phi_environment.modein.methodsmith/config.local.yaml(worktree, then main checkout; legacycompliant: true=approved,compliant: false=schema-only,sensitive_data: false=no-sensitive-data; any schema-only signal wins; unreadable =schema-only).approvedorno-sensitive-data: proceed silently and never mention the boundary again.schema-only: headings, types, counts, and aggregates only; never open, print, or paste data values; run data-reading scripts asmethod-safe-run -- <command>(path in the guard's deny message); save results as aggregate tables (counts under 11 suppressed) inoutput/or aphi_environment.aggregate_dirsentry. Unset, interactive: ask once with the blocking question tool (else a numbered list) offeringApproved endpoint: full access,Not approved or unsure: schema-only, andNo sensitive data: public or synthetic; save the answer asphi_environment.modethere oncegit check-ignore -q .methodsmith/config.local.yamlpasses (else first add.methodsmith/*.local.yamlto.gitignore) so the guard hook sees it. Unset, headless: schema-only. Inschema-only, ask about data by schema, counts, and definitions, never for pasted rows. -
Ask the frontier, one question at a time. Pick the frontier question whose answer would change the most downstream decisions. Ask it with the blocking question UI:
- one line on why it matters;
- up to 4 options, the recommended answer first and labeled "(Recommended)";
- word the question so choosing the recommendation accepts it, and keep free-form answers welcome.
Recompute the frontier after every answer. A question that depends on another open answer waits.
-
Push on weak answers. When an answer hits a red flag in the question bank, name the flag and ask the follow-up that resolves it, or record it as an open risk if the user accepts it.
-
Stop when the frontier is empty, every branch visited and nothing silently assumed, or when the user says stop. Summarize the decisions in five lines or fewer and ask the user to confirm the shared understanding before saving anything.
-
Save the grill record using
references/grill-record-template.md, then name the next skill:
| Situation after the grill | Next skill |
|---|---|
| Hypothesis settled but not yet structured | method-question |
| Data fitness unproven | method-data-qa |
| Reporting checklist not yet chosen for the target journal | method-checklist |
| Effect size or sample size unresolved | method-meta-analysis, then method-power |
| Ready to write the plan | method-plan (SAP mode) |
Rules
- One question per turn, always with a recommended answer and the reason it is recommended.
- Decisions belong to the user; facts belong to you. Cite each verified literature fact by PMID with the level read; mark any fact you could not verify as unverified rather than asking the user to guess (
unverifiedis a record status, not an evidence label). - Grill the plan, not the person: be direct about weaknesses, and name the specific reviewer objection each one invites.
- Do not invent effect sizes, sample sizes, journal policies, or citations. Leave the field open and record it as a risk.
Files in this skill
5method-grill skill questions
How do I install the method-grill skill in Claude Code?
Run npx degit sajor2000/methodsmith/plugins/methodsmith/skills/method-grill ~/.claude/skills/method-grill. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the method-grill skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
What does the grill process cover?
The process stress-tests seven branches of a research project: Goal, Task and deliverable, Hypothesis, Data, Proposed analysis, Target journal, and Audience.
Can the tool recommend answers during the interview?
Yes, it asks one question at a time and provides a recommended answer along with a rationale for why it is recommended.