Search
Glossary8
Open the full glossary- PromptThe entire token sequence sent to the model this turn: system instructions, conversation history, tool definitions, retrieved material, plus the sentence you actually typed — all of it.
- Prompt CachingPrompt Caching reuses a previously cached, byte-identical Prompt prefix to reduce repeated processing; similar but different content is a cache miss.
- Prompt injectionAn attacker hides instructions inside content the AI will read, so the AI treats data as commands.
- System promptA standing instruction written by the developer — not the user — placed at the front of the prompt to frame the model's role, tone and boundaries.
- Prefill and decodeThe two phases of a generation: prefill digests the entire prompt in one parallel pass (compute-bound, sets time-to-first-token); decode then emits one token at a time (memory-bandwidth-bound, sets tokens per second).
- Zero-shot / One-shot / Few-shotThese names only count how many demonstrations the prompt carries: none, one, or a few.
- MCP (Model Context Protocol)An open standard for exposing tools, resources and prompt templates to AI clients — write the server once, and every MCP-compatible client can use it.
- One-off PromptA one-off prompt serves only the task at hand; rules that must hold every time belong in project instructions or skills.
Found 22 articles
AI Supply Chain Poisoning: When the Risk Hides in Models, Data, and Tools
An AI system's supply chain is far more complex than traditional software: model weights, training data, dependencies, external tools (MCP), and even prompts themselves can be tampered with. This article breaks down five links, why detection is so hard, and five practical defensive principles.
Honeypots and Anti-Agent Traps: When the Prey Starts Setting Traps
Honeypots are a classic defensive tool, but once attackers became AI agents both sides changed: defenders now design traps aimed specifically at AI, and anyone using agents to read external content needs to know how to avoid them. This article unpacks four trap classes and how to defend against them.
Build Your First AI Agent in Seven Steps: from environment to a minimal production pass
Adapted from the MIT-licensed awesome-agentic-ai-zh integrated tutorial: one 'paper assistant' project wires up environment setup, your first LLM call, the four-part prompt, tool use, the agent loop with reflection, RAG memory, evals with observability, and the smallest interface. Every step ships minimal runnable code and safety rails; the local Ollama path costs 0 in API fees.
The Evolution of the Harness: Six Eras, Each With Its New Capabilities and New Ways to Crash
From pure-prompt chatbots, CoT and ReAct reasoning prompts, and function calling — the real turning point — to MCP standardization, long-running CLI Agents, and always-on multi-Agent orchestration: era by era, what new capability the harness gained each time, and what new failure mode that capability incidentally introduced.
Compliance Boundaries for Investment-Research AI: Disclaimers, Audit Trails and Traceability
When you put AI into an investment-research workflow, compliance is not one disclaimer line at the end of a report — it is a system-design problem: does the output constitute investment advice, are the data sources licensed, and can every conclusion be traced back to its inputs and prompt versions. This article gives a risk-dimension inventory, an audit-trail architecture, human-review checkpoint design, a question list to take to legal, and a route for strengthening internal policy.
What Is a Terminal: The Name of Everything in That Window You Just Opened
Every install guide's first line is 'open a Terminal.' You open it, see a dark window and a line of text, and don't know where to look next. This isn't a command list — it dissects the screen in front of you line by line: what that window is, why typing can make things happen, what the prompt is telling you, and the invisible but always-active 'current directory.'
Your First AI Task: How to Give Instructions It Gets Right
Why first tasks usually fail — not because the tool is weak, but because the instruction has no output, no input, and no destination. Three elements, three real examples (each with its vague version for contrast), and how to judge how much freedom to give.
Building Organizational AI Capability: From Individual Heroes to Repeatable Processes
A few people in the company being great at using AI does not mean the organization has AI capability. This article is about turning personal experience into organizational assets: capability inventory, seed users and a three-stage rollout path, consolidating personal prompts into a skill library and a template library, tiered training and documentation, and the dimensions and common failure modes for measuring whether it really landed.
Build Your First AI Workflow Without Writing Code
You can build a usable AI application with your own hands even if you cannot write code. Using a customer-support knowledge-base Q&A as the scenario and no-code platforms like Dify, Coze, and n8n, this article walks you through six steps — prepare the data, build the knowledge base, set the prompt, test, launch, monitor — each concrete enough to follow, without depending on interface details that go stale.
The Marketing Content Mass-Production Pipeline: From Topic Selection to Finished Draft
Anyone can write one piece of content with AI; the hard part is stable mass production. This article breaks marketing content production into a pipeline: where topics come from, how to use a style guide as the system prompt to lock in style, how to derive versions for multiple platforms, where to place human review, how to troubleshoot unstable quality, and how to measure whether the line is really producing.
Harness Engineering: The Workbench Around Your Agent
Your agent writes code in demos but breaks in production because you trained a brain without building a workbench. Five diagrams and everyday metaphors unpack Harness Engineering core concepts and rollout steps.
What Is Vibe Coding: Fast, but Fragile
Vibe coding turns plain-language prompts into a working prototype in minutes. So why does it collapse in production? Three failure modes and five guardrails, so you can move fast without breaking things.
NVIDIA SkillSpector: An AI Agent Skill Security Scanner
GitHub Trending #11 · Open-sourced by NVIDIA · Detects 64 vulnerability patterns × 16 risk categories, scanning Agent Skills for malicious code, prompt injection, data exfiltration, and other security risks. A must-have security check before installation
Comparison of AI Agent Verification Architectures: From Prompt Self-Awareness to Architectural Enforcement, a Source-Level Analysis of Seven Approaches
Who verifies that every step of an AI Agent is correct? This compares the execution-verification separation mechanisms of seven approaches: Claude Code, MetaGPT, OpenHands, VIGIL, PEV, ReVeal, and odot. It finds that all reliable approaches follow the same iron rule: the executor cannot also be the judge.
Technical Argument for the OpenClaw Loop Engineering Refactor: A Complete Plan from Prompt-Driven to Architecturally Enforced
Based on source-level analysis of 14 AI Agent architectures and a production environment field audit, this proposes 8 hypotheses for refactoring the OpenClaw system, a comparison of technology options across 8 architecture nodes, and a complete refactor roadmap.
Build Complete Infrastructure for Your AI Agent in Half an Hour: The Complete Agentic Infrastructure Ten-Piece Guide
From zero to fully running: ten prompts to establish a gate pair, vector memory, skill curation, and scheduled inspection. Solves the skill-skipping problem caused by LLM confidence bias, so your Agent never loses its memory again.
200+ Skills, One Router: Engineering Practice of Agent Skill Routing
After installing 200+ skills, does the AI Agent actually get dumber? How category-by-stage matrix routing raises skill discovery from 35% to 90%, reduces incorrect tool usage by 80%, and lets any task automatically match the right skill combination.
AI Security and Red Teaming: Prompt Injection, Jailbreak Defense
A full landscape of security threats to AI systems: Prompt Injection, Jailbreak, Data Poisoning, and practical defense strategies and red team testing methods.
MCP (Model Context Protocol) Ecosystem Explained: The Three Core Primitives of Tool, Resource, and Prompt
A complete guide to Anthropic's MCP protocol: the design principles behind Tools, Resources, and Prompt Templates, practical configuration, and Claude Code integration.
Prompt Engineering Systematic Methodology: From Few-shot to ReAct
A complete methodology for Prompt Engineering: principles and practical applications of techniques such as Few-shot, Chain-of-Thought, Tree-of-Thought, ReAct, and Self-Consistency.
Open Design Hatch Pet Skill Guide
Create Codex compatible animated sprite pets from character design drafts, screenshots, and generated images: 8×9 atlas, transparent unused cells, per-row animation prompts, QA contact sheet, preview video, and pet.json packaging.
Highly Useful Claude Code System Prompt Templates
A curated set of battle-tested Claude Code system prompts, covering common scenarios such as code review, refactoring, and documentation generation.