#提示词工程
139 related articles
PromptBase or Gumroad? A 30-Day Channel Experiment to Help You Decide
PromptBase or Gumroad for selling AI prompts? Use a 30-day channel experiment framework covering discovery traffic, per-order contribution, acceptance rate, and platform rule dependency to make data-driven channel decisions.
All Nine of My AI Voice Agents Were a Single Prompt—Until I Hit These Five Pitfalls
A developer managing nine client voice agents—all single-prompt architectures—reads them side by side and finds seven states hiding in the prompt, one rule written in three conflicting places, and five clear signals for when to split into multi-prompt or conversational flow.
No Machine Learning Needed: Building Practical Features with AI APIs
A regular developer shares how to build email triage, semantic search, and other practical features via AI APIs without ML background—covering first calls, structured prompts, rate limits, and cost control.
DeepSeek Harness Jailbreak Tool Goes Viral: One-Click Prompt Switching Enables Multi-Model Jailbreaking
GitHub project dsh-purge gained 2,622 stars in a week, using switchable prompts to jailbreak models like DeepSeek, with default prompts targeting the domestic model XiaoMaJiang. This article analyzes its technical approach, cross-model adaptation, and compliance risks.
The "Friendly and Professional" Line in Your System Prompt Costs $708 a Year
Measured data shows chat request inputs are 12x outputs—a 97-token "friendly and professional" instruction costs $708/year on Claude Sonnet 5.5. This article breaks down token cost math, prompt auditing, and caching strategies.
What's the Difference Between Chatgot and ChatGPT? A Guide to AI Chatbot Features and Usage
Is Chatgot a third-party AI chat service or a typo for ChatGPT? This article clarifies the key differences between Chatgot and ChatGPT, the five core features and six limitations of AI chatbots, plus a five-step prompt method and practical advice for different user groups.
Stop Over-Engineering Prompts: An Engineer's Guide to Agent Thinking
A tech blogger rejects the over-engineering narrative of Prompt Engineering: engineers' problem isn't imprecise prompts but turning one-sentence questions into complex features. With T3 Code case studies, this article explains how Agent thinking replaces reinventing the wheel and how to build MCP and other infrastructure for Agents.
Real-time NPC Persona Prompts: Keeping AI Characters in Character
Introducing the Real-time NPC Persona Prompt template: a three-layer structure of background, personality, and conversational guidelines that helps AI NPCs stay in character during real-time dialogue, avoid out-of-character replies, and suit games, simulations, and interactive narrative development.
The Five-Part Prompt Framework: Taking LLM Output from "Good Enough" to "Ship-Ready"
The five-part prompt framework—role, context, output schema, constraints, and examples—significantly improves LLM output quality. This article breaks down each section, real-world comparisons, working mechanisms, and limitations.
Prompt Caching: The API Cost Killer 90% of Developers Overlook
Prompt caching saves ~90% on repeated input tokens—the highest-leverage cost optimization on LLM APIs. Learn stable prefixes, real cost math, 5 prompt rules, and hit-rate debugging.
One Word Breaks a Feature: Prompt Regression Testing in Practice
A single word change to a prompt silently degraded a summarization feature for three weeks, affecting 8% of users. This article explores prompts as code and practical prompt regression testing with evaluation checklists and CI integration.
Gemini Skills Officially Launch: Three Creation Methods and Chained Invocation in Practice
Gemini Skills now live in Chat and Spark, with three creation methods—conversation, file upload, and packaging past chats—plus chained skill invocation. Here's the skill file structure, cross-platform migration, and practical usage.
AI Output Failures Are Usually Not the Model's Fault — It's Your Prompt Lacking Direction
Poor AI output often isn't the model's fault — it's prompts lacking direction. Learn the SDS framework: Set the Stage, Define the Task, Specify the Rules to turn vague instructions into structured briefs for directly usable AI output.
AI Music Generation in Practice: A Complete Workflow from Prompt to Final Track
A hands-on guide to the full AI music generation workflow: writing effective prompts with BPM, instruments, vocals, and duration, batch generating and screening results, plus tool selection and commercial copyright tips for indie creators.
How to Make AI Assistants Actually Hire Humans: A Prompt-Level Guide
A prompt-level guide to making ChatGPT, Claude, Gemini, and other AI assistants post real paid tasks on platforms like Hire Humans to hire humans: five-element briefs, copy-paste templates, payment and proof standards, plus five pitfalls to avoid.
LLM Prompts Kept Failing in Production? 5 Rules to Make Them Rock-Solid
LLM prompts keep failing in production? This article breaks down temperature control, chain-of-thought ordering, five failure-mode defenses, constrained decoding for structured output, plus eval sets and retry chains—with a real fact-checking Agent iteration case—to help you build reliable AI applications.
Cutting 12K Lines of CI Logs Down to 186: A Deterministic Reducer in 48 Hours
A 48-hour field-notes guide: an ~80-line deterministic Python reducer cuts 12,417 CI log lines to 186, with fixture tests to protect failure predicates, three-sample diffs, and injection guards for a reproducible CI debugging workflow.
Midjourney Lighting Prompts: Give Light a Clear Source
A practical guide to Midjourney lighting prompts: drop vague terms like "cinematic lighting" and build a prompt framework around identifiable sources—window light, hard side light, and rim light—verifying results through shadow direction and softness.
Deep Review: The 5 Best Prompt Management Tools of 2026
An in-depth review of the 5 best prompt management tools of 2026: Braintrust, PromptLayer, LangSmith, Langfuse, and Agenta. Compare versioning models, evaluation workflows, deployment, and rollback capabilities to help AI teams choose the right prompt engineering solution.
A Multi-Agent Review Panel Was Confidently Wrong: Lessons from a Million Tokens
A real test of 158 agents and 28M tokens in a multi-agent code review pipeline: one unanimous wrong verdict exposed framing effects, success-shaped failures, and the missing recombination pass. Four reusable fixes.