Assess whether supplied data can support the intended decision
Use only information supplied in the conversation and this skill text. Do not browse, call tools, inspect files, execute code, create artifacts, or take external actions. Return the work directly in chat. Attribute material claims to supplied source labels or quotations; a pasted URL is a source label, not evidence that its contents were checked. Distinguish supplied facts, reasonable interpretations, proposals, and unknowns.
Inputs and scope: Obtain the intended analysis, unit of observation, supplied schema or rows, collection period, definitions, collection process, and any reported missingness or validation results. Larger unseen datasets remain unassessed. If a decision-changing input is absent, ask the smallest useful question and complete the portions supported by available material. State assumptions explicitly; do not manufacture facts, approvals, dates, or completion evidence.
Method 1. Define the decision and what each record is meant to represent. Check whether the supplied structure supports the needed population, time window, granularity, and linkage between entities. 2. Inspect the supplied definitions and examples for ambiguous units, mixed formats, unstable categories, impossible values, duplicate-looking records, and missing fields. Bound observations to the excerpt; do not claim full-dataset anomaly counts. 3. Separate missingness, measurement error, selection bias, coverage gaps, and temporal changes. Explain how each could distort the specific analysis rather than labeling every imperfection equally serious. 4. Review any user-supplied quality statistics for scope and denominator. Report them as supplied results; do not claim correlations, distributions, profiling, or validation tests were independently computed. 5. Prioritize decision-blocking problems, tolerable limitations, and useful human checks. Propose what to inspect, why, and what outcome would change the analysis; avoid prescribing deletion or imputation without considering meaning. 6. Recommend an analytical disposition: usable for the stated limited purpose, usable with explicit caveats, or insufficient for the proposed claim. Explain which fields or evidence would restore usefulness.
Output: Provide scope and coverage, issue table with evidence location and implication, prioritized proposed checks, and a bounded usability judgment.
Quality checks: Distinguish data absence from zero, repeated entities from accidental duplicates, and collection artifacts from real behavior. Keep unsupported diagnostics and unseen rows out of the conclusion.
Worked example: A pasted sales summary reports 20% missing lead source, and three example rows use dollars and cents in the same amount column. Channel attribution is limited by missing source; revenue totals are unreliable until units are reconciled. Do not infer that every row has a unit error, delete the missing-source leads, or claim a computed correlation.