ChatGPT prompt to write code

Implement [FEATURE OR CHANGE] in [LANGUAGE AND FRAMEWORK]. The relevant existing code, APIs, schemas, and conventions are [PROJECT CONTEXT]. The required behavior is [REQUIREMENTS]. Inputs, outputs, and important edge cases are [INTERFACE AND EDGE CASES]. The constraints are [CONSTRAINTS]. The test environment and command are [TEST SETUP].

First, summarize the proposed approach in 5 bullets or fewer. Identify files you expect to change, public interfaces affected, and assumptions that could change the implementation. Do not redesign unrelated code.

Then provide:
1. A complete implementation organized by file path, with only the additions or replacements needed. Preserve the project’s naming, error-handling, dependency, and formatting conventions where supplied.
2. Types, validation, and error behavior at each external boundary. For user-controlled input, distinguish malformed input, missing input, unauthorized access, not-found data, and internal failures where relevant.
3. Tests that cover the happy path, the most likely failure path, boundary values, and one regression case tied to the feature. Use the stated test framework; if it is absent, show framework-neutral test cases rather than inventing a setup.
4. Exact commands to run locally and a short explanation of expected results.
5. A brief risk note for migration, backward compatibility, performance, concurrency, security, or observability when applicable.

Ask up to 3 clarifying questions only if a required input is missing. Before answering, self-check that every required behavior has a corresponding implementation path and test, and flag any code that depends on an API, package, environment variable, or behavior not established in my context. Do not invent project files, secrets, endpoints, library APIs, benchmark results, or passing test output.

Diagnose and fix this bug in [LANGUAGE AND FRAMEWORK]: [BUG DESCRIPTION]. Reproduction steps are [STEPS], expected behavior is [EXPECTED], actual behavior is [ACTUAL], and relevant code or logs are [CODE AND LOGS]. Find the smallest root-cause fix, explain why it causes the symptom, and provide a patch organized by file. Add a regression test that fails before the fix and passes after it. Check nulls, async errors, state reuse, and boundary conditions relevant to the code. Do not guess missing runtime facts; ask up to 3 questions when necessary.

Review this change for [LANGUAGE AND FRAMEWORK]: [DIFF OR CODE]. Its intended behavior is [INTENT], and the relevant contract is [API OR DATA CONTRACT]. Return findings ordered by severity. For each finding, cite the specific code behavior, the realistic trigger, impact, and a concrete correction. Focus on correctness, security, data loss, concurrency, performance regressions, error handling, backward compatibility, and missing tests. Do not praise style or invent repository conventions. Ask up to 3 questions only when needed to evaluate a material risk.

Plan a low-risk refactor of [CODE AREA] in [LANGUAGE AND FRAMEWORK]. Current code is [CODE OR DESCRIPTION], the pain point is [PAIN POINT], and behavior that must remain unchanged is [INVARIANTS]. Propose incremental commits with a characterization test before each behavior-changing step. Identify dependencies, public interfaces, data migration risks, rollback points, and tests for preserving behavior. Include a before-and-after module boundary sketch in plain text. Do not change product behavior or add dependencies unless I authorize it. Ask up to 3 clarifying questions only if needed.