Tag
Before anything ships, it helps to have a second pair of eyes. These prompts turn ChatGPT or Claude into that reviewer, flagging issues and ranking them by how much they matter.
25 prompts · page 1 of 3
Audits a dataset's health before analysis starts, using its columns, row count, source, time range, and a sample of rows to spot missing values, type mismatches, likely outliers, duplicate rows, and inconsistent categorical entries. Returns a 1-10 quality score with reasoning and a concrete cleaning step for every issue found, making it a natural first pass before writing any cleaning script.
Builds a repeatable validation suite for a dataset you'll keep receiving, based on its description, columns, and stated business rules. Produces per-column checks, cross-column business-logic rules, and statistical sanity checks, then wraps them into a Python script that reports failure counts per rule, sample bad rows, an overall quality score, and a priority order for fixes.
Documents a dataset column by column for a team about to receive it, using its name, source system, refresh frequency, and column list as inputs. - Per column: type, business meaning, source, valid values or range, null handling - Also covers: dependencies, known quality issues, and 3-5 example values - Output: a clean markdown table plus a JSON schema ready for automated validation
Designs a deduplication pipeline for records — customers, transactions, or products — that carry small formatting or spelling variations instead of exact duplicates, using the dataset description, key identifying fields, and known variation types as inputs. Combines exact-match removal with fuzzy techniques like Levenshtein or phonetic matching, assigns a confidence score to each match, proposes a merge strategy, and generates a human-reviewable report before anything is deleted.
Paste a diff or code block and get a senior-engineer-style first pass before it reaches a human reviewer. - Flags bugs, security risks, performance issues, and maintainability problems - Rates each finding Critical, High, Medium, or Low with line references and a fix - Add project conventions (e.g. "no nested ifs") for sharper, more relevant results - Run it on every PR so reviewers only spend time on what's left after the obvious stuff is caught
Paste any AI-generated code snippet you're about to merge and get a plain-language walkthrough of what it does, how it works step by step, and where its assumptions or edge cases might fail. Built for the moment right after a code-generation tool hands you something you don't fully trust yet, so you catch problems before they reach review instead of during an incident.
Name an algorithm and a target language to get a complete, ready-to-run implementation rather than just the core loop. - Typed main function with clear parameters and return values - Helper functions where the logic benefits from decomposition - Time and space complexity analysis - A runnable example call to verify it works
Give it the PR's stated goal and your repo's architecture and conventions to get a review that avoids both rubber-stamping and nitpicking. It checks correctness, readability, maintainability, test coverage, and backward compatibility in one pass, then sorts findings into Confirmed Issues, Possible Risks, and Nice-to-Haves, closing with three ready-to-paste GitHub review comments.
Describe the functionality, language, and whether you want a class or module to get a clean, well-encapsulated starting point instead of a flat script. - Constructor or initialization with a clear signature - Public methods with docstrings, plus private helpers where useful - Proper encapsulation following OOP principles - Ask for unit test stubs alongside it to lock in the interface early