Prompts for coding agent.
47 prompts · 21–40
This prompt takes a project's files and change instructions as JSON and returns the code to build the requested feature. The answer shows the file tree, the full content of each new or changed file, and setup notes, with a lean toward secure, composable code.
Add JSDoc comments to undocumented functions in a module.
Guide a JSON structure conversion with field mapping, type conversions, phased code, and validation checks.
This prompt turns notes, an idea, or other text into a Mermaid chart that displays in GitHub Markdown. A short plain explanation follows the diagram.
Generate defensive parsing code that validates messy JSON, CSV, XML, or API data into clean structures.
Generate a detailed mobile wireframe description from a product brief.
Translate an analytics request into SQL using only the provided customers, orders, and related tables.
Generate a single HTML file with embedded CSS and JavaScript from a one-page website brief.
Review Python code, identify bottlenecks, and recommend concrete performance optimizations with explanations.
Turn raw, unstructured logs into structured fields and summary statistics.
Generate a production-ready class or module with validation, clear interfaces, and clean API design.
Write a pull request description from a change summary.
Implement a Python function from a natural-language specification with validation, edge-case handling, and comments.
Refactor a messy Python function while preserving behavior.
Generate a regex from a plain-English description with test cases.
Build CRUD operations with the Repository Pattern, validation, and error handling.
This prompt reads the output of terraform plan and writes a Markdown summary. It looks at what infrastructure will change, security risks, likely cost effects, and compliance points, opening with a one-sentence summary.
Design layered client and server validation with schemas, sanitization, error handling, and attack-vector notes.
Generate seed SQL inserts from a database schema.
This prompt designs SLIs and SLOs for a service based on what users experience. It adds an error budget, alerts on how fast it burns, and a policy for when the budget runs out, all in a fixed YAML layout.
A prompt is the text you give a language model so it generates a useful response — instructions, examples, constraints, and your content. Prompt engineering is the craft of writing those instructions clearly. This directory hosts ready-to-copy prompts on the page.
Open a prompt, copy the text, paste it into Claude, ChatGPT, Grok, Gemini, or another chat, fill the placeholders, and send. No account is required on this site.
A prompt is copyable chat text. A skill is an agent pack, often a SKILL.md folder. A method is one way to solve a use case on Can AI? — it might be a prompt, skill, agent, tool, or stack. The same prompt can appear here and as a method on a use case.
Yes. The prompt text is on the page. If we show a source credit, keep that credit. There is no paywall in front of the text.
Someone submits a prompt, or it arrives as part of a method on a use case. An editor publishes it. There is no paid placement.
Standalone prompts for writing, research, coding, ops, and other outcomes, tagged with the same sitewide categories as the rest of the directory so you can browse by area.