question-framing

A planning step for defining an analysis question, metric, or data-based deliverable before working with data. It focuses on the decision the result should support and the exact quantity to measure.

In plain words
What is it for?
It is for framing trend analyses, metrics, models, causal studies, dashboards, charts, maps, tables, and other work built from data.
Why use it?
It prevents producing a correct calculation that answers the wrong question or does not help anyone decide what to do.

Skill for Claude CodeCodex

Part of the causal-powers plugin — 18 skills, 2 agents, 5 hooks shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/lancegui/causal-powers/question-framing
Any agent
npx skills add lancegui/causal-powers --skill question-framing
Clone the repo
git clone --depth 1 https://github.com/lancegui/causal-powers

Made for: Claude Code, Codex.

Or install causal-powers, the plugin that ships this one along with the rest of its 18 skills, 2 agents, 5 hooks.

Per session 214 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,123 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00214 $0.03123
Opus 5 $0.00107 $0.01562
Sonnet 5 $0.00043 $0.00625
Haiku 4.5 $0.00021 $0.00312

Measured 2d ago against content hash a25b7fb0d2e2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

question-framing scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/question-framing/SKILL.md · 120 lines

How it starts

The opening of the file, as written. The whole thing — 120 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Question Framing

Overview

The most expensive analytics mistake is not a wrong number — it's a right answer to the wrong question. It survives every validation check, reconciles perfectly, reproduces exactly, and is still useless, because the metric measured something other than what the decision needed.

This is the analytics counterpart of brainstorming a feature before building it — nail down what you're actually being asked, and what a good answer would change, before you load data.

Core principle: Define the estimand and the decision before you touch the data — because once you see the data, your definition will quietly bend to fit what's easy to compute.

The framing brief

Produce a short, complete brief — not a sprawling document — answering the elements below; for general/exploratory work it also fixes the data, approach, and deliverable (The plan below). Each one is a place analyses go wrong:

  1. The decision. What action does this number inform, and who takes it? If no decision rides on it, scope it down or drop it. "Interesting" is not a spec.
  2. The estimand / metric, exactly. Not "engagement" but "median sessions per 7-day-active user, per calendar week, in the US." Not "the effect of the pricing change" but "the change in 30-day retention for users who saw the new price vs. those who didn't." Pin the numerator, denominator, unit, and time window.
  3. Population and filters. Who is in and who is out? New vs. existing? Which date range? Which segments? Every filter is an assumption — name it.
  4. Unit of observation. Per user? per session? per transaction? per user-week? Most double-counting and most wrong denominators trace to a fuzzy unit of analysis.
  5. What would change the answer / decision. What result would flip the decision? If any number leads to the same action, you don't need the analysis. This also tells you the precision you actually need.

For a causal question, add three more and hand off to causal-identification:

Read the full file on GitHub · 120 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 120 lines · 214 tokens per session scan A a25b7fb0d2e2

Subscribe to this mod's changes

question-framing is a skill published in the GitHub repository lancegui/causal-powers (2 stars, last pushed 9d ago), licensed MIT. It adds 214 tokens to every session and 3,123 once invoked, about $0.0011 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

audit-reproducibility

Enforce the replication-protocol.md rule by cross-checking numeric claims in a manuscript against the actual R / Stata / Python outputs. Report PASS/FAIL per claim against tolerance thresholds. Use before submission and before releasing a replication package.

pedrohcgs/claude-code-my-workflow · 54 tokens

diagnose

Root-cause a failing or wrong empirical result with a disciplined reproduce → minimise → hypothesise → instrument → fix loop, instead of guessing-and-poking. Use when the user says "why is my regression wrong", "this number changed", "my script errors out", "the result won't reproduce", "debug this", "this estimate…

pedrohcgs/claude-code-my-workflow · 145 tokens

capture-environment

Snapshot the computational environment for a replication package — detects the analysis stack (R / Stata / Python) and emits the right lockfiles (renv.lock + sessionInfo.txt, requirements.txt / environment.yml / uv.lock, Stata version + ado package list), records seeds and RNG kind, optionally writes a pinning…

pedrohcgs/claude-code-my-workflow · 139 tokens

checkpoint

Save a structured state snapshot before stopping or handing off. Captures the active plan, recent decisions, file pointers (with line numbers), open questions, and the next 1–3 actions into a checkpoint file under qualityreports/checkpoints/. Optionally proposes [LEARN] entries to add to MEMORY.md. Use when user says…

pedrohcgs/claude-code-my-workflow · 124 tokens

coauthor-brief

Generate a co-author / collaborator handoff brief for a multi-author, multi-machine project — summarizing what changed since the last brief (git delta), the current state of each artifact (manuscript, analysis, slides), open questions, how to reproduce locally, and any restricted-data access steps. Use when user says…

pedrohcgs/claude-code-my-workflow · 146 tokens

data-management-plan

Draft a funder-compliant Data Management Plan (NSF DMP, NIH DMS Policy 2023, ERC, Horizon Europe) by composing the confidential-data and environment-capture primitives. Sections cover data description, formats/metadata, storage/backup, access/sharing, preservation/archiving, and roles. Use when user says "data…

pedrohcgs/claude-code-my-workflow · 143 tokens