Front page / Concepts / Prompt engineering
Prompt engineering
Crafting the wording of a request so the model answers well: role statements, step by step instructions, examples, output formats. The techniques still work, but as a stand-alone skill it has been absorbed into context engineering, where the wording is one small part of what the model is given. Knowing the patterns helps; collecting prompts does not.
Where it stands
Among the 14 concepts. Bars: GitHub stars gained this week.
- MCP and tool use+3.9k
- Agentic loops+1.3k
- Knowledge graphs (GraphRAG)+1.2k
- Retrieval (RAG)+1.2k
- Prompt engineering+1.1k
- Vibe coding+980
- Workflows, not agents+565
- Spec-driven work+16k
- Memory+7.3k
- Agent skills+5.7k
- Multi-agent orchestration+5.6k
- Computer use+4.7k
- Context engineering+3.3k
- Evals and verification+929
How to use it today
In ChatGPT or Claude, as a person doing the work, not as a developer.
- 1Keep the three patterns that still pay: say who the reader is, give one example of the output, name the format.
- 2Replace a prompt you reuse with a skill or a Project instruction; a prompt you paste every week is a skill you have not installed.
- 3Drop the hype words ("world-class", "CRITICAL", "think step by step"): current models think on their own and overreact to shouting. One plain sentence about the role and the reader still helps.
- 4When an answer disappoints, change what the model was given before changing how you asked.
What changes day to day
You stop hunting for the right phrasing and stop keeping a prompt file. Effort moves to the material and the acceptance criteria.
Where it goes wrong
Prompt collections age fast and encode habits from weaker models.
Chasing phrasing hides the real problem, which is almost always missing or contradictory context.
Listings that use it well
Chosen by hand after reading them, each with the reason. Install from the listing's own page.
| Listing | Kind | Why it is here | Stars |
|---|---|---|---|
| llm-application-dev | Agent skills | prompt-engineering-patterns for optimizing, templating and debugging prompts. | 42k |
| engineering | Agent skills | prompt-governance versions prompts, A/B tests them and prevents regressions in production. | 29k |
| claude-skills | Agent skills | prompt-engineer writes and evaluates prompts with templates, output schemas and rubrics. | 12k |
| prompt-master | Agent skills | Writes optimized prompts for any AI tool. | 16k |
| claude-code-prompt-improver | Agent skills | A hook that rewrites vague Claude Code requests into precise prompts. | 2k |
Elsewhere
Repositories that embody the idea and are not in the catalog.
- dair-ai/Prompt-Engineering-Guide ↗The most-starred prompt engineering guide; its own description now adds context engineering, RAG and agents.
- anthropics/prompt-eng-interactive-tutorial ↗Anthropic's hands-on notebook course on prompt structure and failure modes.
- f/awesome-chatgpt-prompts ↗The prompt-collection era in one repository (since December 2022).
Standing
Our judgment, with the reason and the sources it rests on.
Established since 2025-06. The techniques are alive and the vendors still publish a prompting guide for each new model, but the advice changed: say the outcome, the audience and the limits, and cut the rest. Since mid 2025 the field treats wording as one part of context engineering. Settled practice: the methods are known, the tools are mature, and the question is how well you use it.
- Andrej Karpathy on X: +1 for "context engineering" over "prompt engineering"2025-06-25
- Effective context engineering for AI agents2025-09-29
- Prompt guidance for GPT-5.62026-07
Related: Context engineering, Agent skills