惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

N
Netflix TechBlog - Medium
Blog — PlanetScale
Blog — PlanetScale
月光博客
月光博客
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
CTFtime.org: upcoming CTF events
CTFtime.org: upcoming CTF events
爱范儿
爱范儿
量子位
博客园 - 聂微东
Engineering at Meta
Engineering at Meta
WordPress大学
WordPress大学
GbyAI
GbyAI
MyScale Blog
MyScale Blog
IT之家
IT之家
P
Proofpoint News Feed
M
MIT News - Artificial intelligence
The Cloudflare Blog
H
Hackread – Cybersecurity News, Data Breaches, AI and More
Hugging Face - Blog
Hugging Face - Blog
The Register - Security
The Register - Security
Microsoft Security Blog
Microsoft Security Blog
博客园_首页
MongoDB | Blog
MongoDB | Blog
F
Fortinet All Blogs
博客园 - 三生石上(FineUI控件)
Y
Y Combinator Blog
雷峰网
雷峰网
V
Visual Studio Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
Last Week in AI
Last Week in AI
博客园 - 叶小钗
D
DataBreaches.Net
B
Blog
B
Blog RSS Feed
大猫的无限游戏
大猫的无限游戏
aimingoo的专栏
aimingoo的专栏
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
The GitHub Blog
The GitHub Blog
云风的 BLOG
云风的 BLOG
Recent Announcements
Recent Announcements
阮一峰的网络日志
阮一峰的网络日志
小众软件
小众软件
腾讯CDC
T
Threat Research - Cisco Blogs
SecWiki News
SecWiki News
Martin Fowler
Martin Fowler
D
Docker
Cisco Talos Blog
Cisco Talos Blog
T
Tenable Blog
Webroot Blog
Webroot Blog
宝玉的分享
宝玉的分享

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders How to Switch AI Chatbots and Why You Might Want To GitHub - MattMessinger1/agentic_refund_guardrail: Safe refund policy layer for AI agents — Python + TypeScript. Same behavior, shared tests. Adam/papers/emergent_values_whitepaper.md at master · strangeadvancedmarketing/Adam Ask HN: How do you stop playing 20 questions with your AI coding tools How far can automation and AI support psychotherapy? - @theU GitHub - stagas/rtdiff: realtime git diff gui and AI-assisted commits A Mac Studio for Local AI — 6 Months Later A History of the Early Years of AI at the University of Edinburgh Why AI Coding Tools Still Feel Stuck on Localhost MSN AI Datacenters Are Becoming Strategic Targets twitter.com Penn Researchers Use AI to Surface Unreported GLP-1 Side Effects in Reddit Posts Show HN: MoodSense AI (ML and FastAPI and Gradio, Deployed on Hugging Face) Moodsense Ai - a Hugging Face Space by aman179102 AI models are terrible at betting on soccer—especially xAI Grok GitHub - xialeistudio/echoic GitHub - HimashaHerath/github-dev-wrapped: AI-powered weekly GitHub activity reports deployed to GitHub Pages GitHub - alejandrobalderas/claude-code-from-source: Architecture, patterns & internals of Anthropic's AI coding agent — reverse-engineered from source maps AI and Tech brief: Ireland ascendant GitHub - Titovilal/context0: Context0 - Never Surrender Training for a Marathon with an AI Coach: What Worked and What Didn't Cyber Pulse: Agentic Intel - Apps on Google Play I Built an AI PR Reviewer That Catches Bugs by Not Looking for Bugs Gen Z workers are so fearful AI will take their job they’re intentionally sabotaging their company’s AI rollout | Fortune How AI Is Reimagining the Game of Golf–For Both Players and Courses GitHub - nattergabriel/reseed: A CLI tool for managing and distributing agent skills across projects Is SVG the final frontier? My AI workflow evolved from prompts to a near-autonomous workflow MLSharp Help - 3DGS Viewer & Generator I put my cognitive field based AI's runtime on GitHub Is Numble the first AI-proof game? A3: Kubernetes for autonomous AI agent fleets | Emergent Principles Deepali Vyas ("The Elite Recruiter") GitHub - msmarkgu/RelayFreeLLM: A restful API designed to route user prompts to various AI model providers. Unionized ProPublica staff are on strike over AI, layoffs, and wages Unleashing the Advantage of Quantum AI We're heading for an AI-fueled 'dementia crisis,' brain scientist warns The AI-Assisted Breach of Mexico's Government Infrastructure [pdf] GitHub - stef41/lmscan: 🔍 Detect AI-generated text and fingerprint which LLM wrote it. Open-source GPTZero alternative. Zero dependencies, works offline. MSN GitHub - visionscaper/collabmem: Enabling long-term collaboration with Agentic AI - building up episodic and world model memory over time with in-context awareness We gave an AI a 3 year retail lease in SF and asked it to make a profit | Andon Labs AI Code is Hollowing Out Open Source, and Maintainers are Looking the Other Way What leaked "SteamGPT" files could mean for the PC gaming platform's use of AI AI is the boss at this retail store. What could go wrong? GitHub - Wuzu11517/agentic-proxy: Local proxy meant to help reduce With Drones, Geophysics and ArtificiaI Intelligence, Researchers Prepare to Do Battle Against Land Mines A Single Operator, Two AI Platforms, Nine Government Agencies: The Full Technical Report 在 Steam 上购买 FriedrichAI: Offline AI 立省 10% GitHub - inevolin/resume-cli: Hit Claude usage limits? Resume any AI coding session elsewhere. Switch tools at zero friction. GitHub - atripati/ark: AI Runtime Kernel — a context operating system for AI agents. Eliminates tool bloat, loads only what’s needed, and gives LLMs their reasoning space back. How to Build a Secure AI PR Reviewer with Claude, GitHub Actions, and JavaScript This Startup Wants You to Pay Up to Talk With AI Versions of Human Experts Intel Arc Pro B70 Brings 32GB VRAM to Local AI for $949 WordPress 7.0: The Good, the AI, and the Still Missing AI on the couch: Anthropic gives Claude 20 hours of psychiatry IatroBench: Pre-Registered Evidence of Iatrogenic Harm from AI Safety Measures AI Agents Know About Supabase. They Don't Always Use It Right. The history and future of AI at Google, with Sundar Pichai Inside an AI‑enabled device code phishing campaign How Meta Used AI to Map Tribal Knowledge in Large-Scale Data Pipelines AI for Systems: Using LLMs to Optimize Database Query Execution Forecasting the Economic Effects of AI Introducing Tinker: Play with AI, bring your ideas to life AI sheds light on an ancient gaming mystery People really hate AI but not as much as Iran—or Democrats | Fortune What is an AI Product Engineer? Phoebe Gates wants her $185 million AI startup to succeed with 'no ties to my privilege or my last name': 'I have a chip on my shoulder' | Fortune
Prompt Foundry - Visual Studio Marketplace
espresso_dev · 2026-06-19 · via Hacker News - Newest: "AI"

Jump to: Demo | Setup instructions

Prompt Foundry is a VS Code extension that makes AI more consistent and effective in large codebases. Instead of rewriting context every session, you build a personal library of reusable prompt blocks* — architecture notes, coding standards, behavioural contracts — and snap them together before each task. A bundled MCP server lets the AI write discoveries back into that library, so your prompts improve over time without manual upkeep.

*A prompt block a markdown file that represents template prompt in your library which is compiled with other blocks and your main instruction to form a complete prompt before sending to the AI.

Problem How Foundry handles it
I've added so many agents.md instructions and skills, the AI gets confused Task-specific prompt blocks keep global context lean and targeted
Prompt library not accessible from Claude Code CLI TUI to attach instructions to your prompt directly from the CLI tool
AI doesn't behave how I want it AI Contract defines role, style, and behavioural expectations upfront
Some instructions matter more than others Star a prompt block to surface it as a key goal, or set a reference to position it at the right point in the workflow
My MCP tool to my local file system is not powerful enough Use Liquid templating syntax to make prompt blocks more specific; use MCP tools to append to blocks (e.g. keep a log as the AI goes); review changes before committing to the prompt library
Knowledge from AI sessions gets lost Maintain a personal library of context and architecture files (second brain). Updated via MCP tool: "please update xxx block with what we learnt"

Overview of features

VS Code extension MCP server (optional) TUI (optional)
VS Code extension MCP TUI
Manage prompt library, compile prompts, @ mentions and add selection from editor settings Give AI access to your prompt block library to update content Attach prompt blocks to your prompt from the command line. E.g. in Claude Code you can press Ctrl+G to open the picker directly

Benchmark

The AI was asked to build a TUI version of Prompt Foundry (this is now a bundled feature!). Run #1 is a baseline prompt. Run #2 uses Prompt Foundry. Both use Gemini Flash Lite 3.1.

The key differences: better code structure and far less handholding.

Run Achieved Task Code Structure Handholding Remaining Sig. Bugs
#1 Baseline Yes 1 large component Yes (had to re-align) Scrolling bug
#2 Foundry Yes 3 components, 1 test Minimal (errors/next steps)

Run #2 also followed instructions in the prompt blocks, generating code according to the defined style.

Make a Plan Used Information TDD Approach Code Comments Added Logs
✓ including all sections ✓ Used to stay on track Attempted ✓ Some added ✓ Logs

Details: Baseline | Foundry run

AI comparison of diffs

Both runs used ink (React-based TUI) and clipboardy. The architectural differences matter most if this TUI is to share a core backend with the VS Code extension, which is the intended direction.

Baseline (#1) added commander and meow — two CLI argument parsers doing the same job. It also added chalk for terminal styling alongside ink, creating two competing approaches: ink's declarative React component model vs. imperative ANSI escape codes. This kind of mixed-paradigm dependency set creates friction as the codebase grows. It also included no test infrastructure.

Run #2 dropped both redundant dependencies, kept styling within ink's component model, extended the component set with ink-select-input, and added jest, ts-jest, and ink-testing-library. The core business logic (LibraryManager, SessionManager, PromptCompiler) already has zero VS Code dependencies by design. Run #2's test setup means TUI components and core logic can both be unit-tested in isolation, without spinning up a VS Code instance or a live terminal.

In short: Run #1 works but accumulates debt. Run #2 reflects an understanding of where the project is heading.

How it works

  1. Type your prompt
  2. Select your prompt blocks (instructions and information)
  3. Mark the most important ones as goals
  4. Copy/send to AI
  5. Close the loop with MCP: At the end of a session, tell the AI to update your library — e.g. *"update the auth-architecture block using the prompt foundry mcp with what we just decided." Next session, that knowledge is already there.

The self-improving library

The MCP server exposes your prompt block library as readable and writable files. During a session the AI can:

  • Read any block to get up-to-date context mid-task
  • Append to a block — useful for keeping a running log of decisions
  • Overwrite a block with revised content, which you review before committing

This means context compounds. Architectural decisions, gotchas, naming conventions — anything worth remembering gets written back in, in the right place, ready for the next session.

Example output prompt structure:

Example prompt

See demo

Setup

Extension

Install the extension from the VS Code Marketplace. This bundles with the MCP server and TUI. The extension includes a default block library to get you started. Blocks are read-only until you create your own prompt library directory — set the location via the gear icon next to prompt blocks or through VS Code settings.

Note: When editing from the editor you need to open the prompt library folder in VS Code and click 'Trust'. The folder only contains .md files and prompt settings files.

MCP server

Note: The MCP server requires Node.js to be installed on your machine.

  1. Open the extension
  2. Click the gear next to Prompt Block Library
  3. Click Features & Permissions > MCP Server
  4. Copy the generated JSON snippet
  5. Add it to your AI's MCP config

Note: The MCP server runs locally as a Node.js process in the VS Code extension folder. The AI can read and modify the content of the specified prompt library folder via MCP tool calls.

TUI

The TUI provides a way to quickly attach prompt blocks to a prompt in another app. For instance, in Claude Code you can press Ctrl+G to open an external editor, which can be configured to use this tool.

  1. Open the extension
  2. Click the gear next to Prompt Block Library
  3. Click TUI Dashboard
  4. Follow the instructions for Claude Code or to set the external editor

Features

Instructions prompt

Enter your main instructional prompt into the top instruction box.

The live focus (⚡) button lets you select files and lines in the IDE and adds those locations to the prompt. Useful for dictating while navigating a codebase — you end up with contextual file tags as you navigate, similar to someone watching your screen as you explain.

Prompt blocks

A prompt block is a Markdown file containing reusable instructions or information. You have your own prompt library but can also attach a custom directory or workspace directory. Here is a simple example:

# Code style

- Use TypeScript strict mode
- Prefer named exports over default exports
- Add JSDoc comments to all public functions

Blocks are organised into categories, one per folder, plus a set of special categories. Optionally add your Claude or Cursor skills too.

References

A block can include a short reference — a reminder compiled into a specific position in the prompt. This controls where in the workflow the AI sees it, rather than dumping all instructions in one place.

referenceLocation controls where in the compiled prompt the reminder appears:

Value Position
workflowFirstTurn Start of the first turn
workflowEveryChange Before every code change
workflowBeforeEditing Before the AI starts editing
workflowEndOfTask End of the task
pre Top of the prompt
remark General remark

Goals

Star a block to mark it as a goal. Its reference text gets pulled into a dedicated # Key goals while completing this task: section at the end of the prompt, giving the AI a clear summary of what matters most before it starts work.

Templating

Prompt blocks support Liquid syntax. Useful when you want to add custom variables to a block without rewriting it:

{% comment %}
vars:
  selectExample:
    type: select
    options: [
      "One",
      "Two",
    ]
  textExample:
   type: text
{% endcomment %}
Select option: {{ selectExample }}

Text example: {{ textExample }}

Special categories

  • AI Contract (editable template): Define role, comment style, and other behavioural expectations. The extension structures the prompt to encourage the AI to stick to the contract.
  • Tools:
    • Git commit: Add a specific git commit.
    • Git diff: Add a diff against a branch or commit hash.
    • IDE diagnostics: Share detailed errors if you don't have IDE MCP set up.
    • Active symbols: Add file summaries for context.
  • Claude skills and commands: Lists Claude skills and commands from your global Claude folder and prompts the AI to use them.
  • Cursor rules: Same as above.

Prompt block groups

When you frequently need to add the same prompt blocks, use the group feature. To create a group, add some blocks to the prompt and then press the '+' button next to group. The active prompts will be saved to the group.

Demo

Prompt Foundry Demo 1
Building a prompt with reusable blocks

Prompt Foundry Demo 2
Using MCP server to update blocks

License & Privacy

License: MIT © Simon de Vries

Privacy: 100% local. No telemetry, no data collection. Your prompts never leave your machine.

Feedback

No telemetry means I rely on you. Suggestions and feature requests welcome: https://form.typeform.com/to/hAc2CQ6A