4 Agents to Spec, Build, Break, and Fix a Feature
Four agents take turns: write the spec, build it, try to break it, then fix it. Why it matters: fewer half-done features.
Coding prompt series
Coding workflow prompts
Prompts for everyday engineering work: debugging, refactors, code review, test suites, migrations, CI pipelines, threat models, and system design.
32 prompts in this series. Each prompt page shows the full prompt, the variables to fill in, the models it was tested with, and example output.
All prompts in this series
32 coding workflow prompts
Agent Report Card (Pass/Fail Tests)
Builds simple pass/fail tests for an AI agent and scores a run. Why it matters: if you can't test it, you can't trust it.
Generate API Test Cases from a Contract
Converts API documentation into a coverage matrix and runnable tests for successful, invalid, and edge-case requests. Why it matters: clear steps stop silent bugs later.
Review an Architecture Diagram for Risks
Examines a system diagram for failure points, trust boundaries, scaling risks, and hidden dependencies. Why it matters: clear steps stop silent bugs later.
REST API Design — ChatGPT Prompt
Design a REST API with resources, errors, auth notes, and examples. Why it matters: clear steps stop silent bugs later.
GitHub Actions CI Pipeline — ChatGPT Prompt
Draft a GitHub Actions workflow for test, lint, and deploy gates. Why it matters: clear steps stop silent bugs later.
Debugging Root Cause Analysis — ChatGPT Prompt
Drive a structured RCA from logs, stack traces, and recent changes. Why it matters: clear steps stop silent bugs later.
Hardened Dockerfile — ChatGPT Prompt
Writes a minimal hardened Dockerfile with multi-stage and non-root user. Why it matters: clear steps stop silent bugs later.
Developer-Friendly Error Messages — ChatGPT Prompt
Rewrite error messages for clarity, actionability, and safe disclosure. Why it matters: clear steps stop silent bugs later.
Blameless Incident Postmortem — ChatGPT Prompt
Draft a blameless postmortem from timeline notes and impact. Why it matters: a reusable format beats starting from zero.
Coding Interview Problem Coach — ChatGPT Prompt
Coach through a coding problem without dumping the full solution first. Why it matters: clear stories beat memorized buzzwords.
Database Migration Plan — ChatGPT Prompt
Plan a zero/low-downtime schema migration with rollback steps. Why it matters: clear steps stop silent bugs later.
Performance Profiling Plan — ChatGPT Prompt
Makes a profiling plan for a slow endpoint or job. Why it matters: clear steps stop silent bugs later.
Python CLI Scaffold — ChatGPT Prompt
Scaffold a Python CLI with argparse/typer, logging, and tests. Why it matters: clear steps stop silent bugs later.
Accessible React Component — ChatGPT Prompt
Implement or review a React component with accessibility checklist. Why it matters: clear steps stop silent bugs later.
Safe Refactor Plan — ChatGPT Prompt
Produce a step-by-step refactor plan with characterization tests first. Why it matters: clear steps stop silent bugs later.
Regex with Tests — ChatGPT Prompt
Build a regex with explanation and positive/negative test cases. Why it matters: clear steps stop silent bugs later.
Pull Request Summary — ChatGPT Prompt
Summarize a PR for reviewers: intent, risk, test plan, and rollout notes. Why it matters: clear steps stop silent bugs later.
Code Review Prompt That Checks If It Actually Works
Reviews code for real bugs and broken cases first — kindness is optional. Why it matters: pretty code that fails still fails.
Security-Focused Code Review — ChatGPT Prompt
Review code for common security issues with severity and fix guidance. Why it matters: clear steps stop silent bugs later.
SQL Query Optimization — ChatGPT Prompt
Optimize a SQL query with EXPLAIN reasoning and index suggestions. Why it matters: clear steps stop silent bugs later.
System Design Interview Practice — ChatGPT Prompt
Run a system design interview with clarifying questions and tradeoffs. Why it matters: clear stories beat memorized buzzwords.
STRIDE Threat Model — ChatGPT Prompt
Produce a lightweight STRIDE threat model for a feature. Why it matters: clear steps stop silent bugs later.
TypeScript Types from JSON — ChatGPT Prompt
Infer careful TypeScript types from sample JSON with validation notes. Why it matters: clear steps stop silent bugs later.
Unit Test Suite from Spec — ChatGPT Prompt
Generate unit tests from a function spec with edge cases and fixtures. Why it matters: clear steps stop silent bugs later.
Explain This Legacy Codebase to Me
Maps entry points, data flow, important modules, side effects, risks, and safe starting tasks. Why it matters: clear steps stop silent bugs later.
Incident Commander Kit
Helps run an outage: status note, customer note, and postmortem outline. Why it matters: panic without a checklist makes outages worse.
Create a Minimal Bug-Reproduction Plan
Turns an issue report, logs, and code context into testable hypotheses and a minimal reproduction strategy. Why it matters: clear steps stop silent bugs later.
Convert Requirements into Acceptance Tests
Generates positive, negative, boundary, permission, accessibility, and recovery scenarios from requirements. Why it matters: clear steps stop silent bugs later.
Plan a Safe Dependency Upgrade
Identifies breaking changes, affected code, migration steps, tests, rollout risks, and rollback criteria. Why it matters: simple language helps kids learn.
Generate SQL with a Validation Plan
Produces SQL together with assumptions, expected row behavior, sanity checks, and performance considerations. Why it matters: clear steps stop silent bugs later.
Turn My Workflow Into a Cursor Skill
Turns how you work into a Cursor skill file others can reuse. Why it matters: good workflows die if they stay in your head.
Coding prompts
More coding prompt series
Changelog from commits prompts
6 prompts · Turn a commit log into a readable changelog and release notes, with language-specific versions.
Docstring generator prompts
6 prompts · Generate idiomatic docstrings and API comments for Go, Java, Python, Ruby, SQL, and Swift code.
Feature flag rollout plan prompts
6 prompts · Plan a feature flag rollout, kill switch, and cleanup, with one prompt per language.
GraphQL schema design prompts
6 prompts · Design a GraphQL schema, resolvers, and pagination for Go, Java, Python, Ruby, SQL, or Swift backends.
Observability checklist prompts
6 prompts · Logging, metrics, tracing, and alerting checklists for a service, with one prompt per language.
Architecture decision record (ADR) prompts
5 prompts · Write an architecture decision record for a C, Kotlin, PHP, Rust, or TypeScript codebase: context, options, decision, and consequences.
RBAC authorization matrix prompts
5 prompts · Design a role-based access control matrix and enforcement plan, with one prompt per language.
Kafka consumer design prompts
5 prompts · Design a Kafka consumer with retries, idempotency, and dead-letter handling, with one prompt per language.
Load test plan prompts
5 prompts · Write a load test plan with scenarios, targets, and pass/fail thresholds for Go, Java, Python, Ruby, or Swift services.
API rate limit design prompts
5 prompts · Design API rate limiting (token bucket, sliding window, quotas, and 429 handling) for Go, Java, Python, SQL, or Swift.