Show HN: Workflow86 - An AI business analyst and automation engineer https://ift.tt/KkF4wNT

Show HN: Workflow86 - An AI business analyst and automation engineer Hey HN, We built Workflow86 to help teams build and automate their internal business processes and workflows using drag and drop components like forms, tasks, tables and nodes for business logic, API requests, running custom code etc. It works as a standalone process/workflow automation tool, or as a workflow customization layer on top of existing apps and systems like HRIS, CRM and ERP. One common problem we hear from users is that no-code still has a significant learning curve, and it can take some time to understand how to properly build something. Users also needed help with knowing what to build in the first place, or what a process might or should look like. To solve this, we've integrated an AI that acts as a business analyst/consultant and workflow automation engineer. This AI is powered by a combination of Large Language Models and lots of prompt engineering, RAG and prompt chaining techniques we developed along the way. See a demo of it in action here: https://ift.tt/lv7LjD0?... In business analyst/consultant mode, the AI helps users brainstorm ideas, identify and discover processes and draft what a process should look like. Like a business analyst/consultant, the AI works to pull and extract information and details from the user by asking the right questions rather than rely on the user's instructions alone. Once the required information has been gathered, the AI goes into engineer mode: it will plan and then build the entire workflow by selecting the right nodes, connecting them together and then fully configuring every single node individually as well. This includes writing custom code and API requests using stored credentials when required. Once a workflow is built, edits can be done manually or by asking the AI to adjust the workflow at any time (e.g., “Add a compensation band check before final approval”). The AI has full context of the current state of the workflow, so it can “patch” in any changes like adding new nodes, rewriting existing nodes and so on. Some use cases we’ve seen from customers include building: - automated compliance checks for new CRM leads - custom international contractor onboarding workflows on top of a HRIS - automated vendor risk assessment before ERP updates Try it out and let us know how the AI performs and any other feedback you have! Full docs can be found at https://ift.tt/A2y8kW6 https://ift.tt/fzOWPq9 January 30, 2025 at 10:35PM

Show HN: Audiocube – A 3D DAW for Spatial Audio https://ift.tt/h4rAivc

Show HN: Audiocube – A 3D DAW for Spatial Audio I’ve recently released my solo project Audiocube I wanted to make a 3D DAW, where spatial audio, physics, and virtual acoustics are all directly integrated into the engine. This makes it easy to create music in 3D, and experiment with new techniques which aren’t possible in traditional DAWs and plugins. I’d love to get any feedback on this software (Mac/Windows) to make it better. You can download it for free through the website. Thanks, Noah https://ift.tt/QjA9ocJ January 30, 2025 at 06:42PM

Show HN: Iterm-Mcp – AI Terminal/REPL Control for iTerm2 https://ift.tt/dgsZ7pj

Show HN: Iterm-Mcp – AI Terminal/REPL Control for iTerm2 Hi HN! Ever wish you could just point your AI assistant at your terminal and say 'what's wrong with this output?' That's why I built iterm-mcp. It lets MCP clients like Claude Desktop directly interact with your iTerm2 terminal - reading logs, running commands, using REPLs, and helping debug issues. Want to explore data or debug using a REPL? The AI can start the REPL, run commands, and help interpret the results. This is an MCP server that integrates with Claude Desktop, LibreChat, and other Model Context Protocol compatible clients. https://ift.tt/8KLXDrf Note: Independent project, not officially affiliated with iTerm2 ## Features *Efficient Token Use:* iterm-mcp gives the model the ability to inspect only the output that the model is interested in. The model typically only wants to see the last few lines of output even for long running commands. *Natural Integration:* You share iTerm with the model. You can ask questions about what's on the screen, or delegate a task to the model and watch as it performs each step. *Full Terminal Control and REPL support:* The model can start and interact with REPL's as well as send control characters like ctrl-c, ctrl-z, etc. *Easy on the Dependencies:* iterm-mcp is built with minimal dependencies and is runnable via npx. It's designed to be easy to add to Claude Desktop and other MCP clients. It should just work. ## Real-World Example: Debugging Sidekiq Jobs I needed to debug a Sidekiq job with complex arguments. The arguments were partially obfuscated in the logs. I asked Claude: "open rails console, show me arguments for the latest XYZ job". The model: 1. Launched Rails console 2. Retrieved job details 3. Displayed the arguments that I was looking for ## Architectural Journey This project had a couple interesting constraints around command execution: ### 1. Token Efficiency Challenge I wanted to constrain tokens as much as possible. I didn't want to send the entire output of a long running command to the model, but there's not a great way to know which parts of the output are important to what the model is doing. Sampling could be used here, but it's not well supported yet. *Solution:* I arrived at a pull-based solution for this. The command from the model is sent to the terminal, and the model is made aware of how many lines of output were generated. The model can choose to retrieve as many lines of the buffer that it thinks are relevant. ### 2. Long-Running Process Support I wanted to support long running processes. It turns out that when you run `brew install ffmpeg` - it takes a while, and it's not always clear when the job is done. In early proof of concepts, the model would assume the command completed successfully and begin sending additional commands to the terminal before the first command had finished. *Solution:* iTerm provides a way to ask if the terminal is waiting for user input, but I found that it tended to show false positives in certain situations. For example, a long running command would result in iTerm reporting that the terminal was waiting for input when in fact the command was still running. I found that inspecting the processes associated with the terminal and waiting until the most interesting of those processes settles to a low resource usage is a fair indicator of long running commands being ready for input. ## Requirements * iTerm2 must be running * Node version 18 or greater ## Safety Considerations * The user is responsible for using the tool safely. * No built-in restrictions: iterm-mcp makes no attempt to evaluate the safety of commands that are executed. * Models can behave in unexpected ways. The user is expected to monitor activity and abort when appropriate. * For multi-step tasks, you may need to interrupt the model if it goes off track. Start with smaller, focused tasks until you're familiar with how the model behaves. https://ift.tt/8KLXDrf January 30, 2025 at 11:44PM

Show HN: I made a price aggregator to find the best PC parts deals across eBay https://ift.tt/9KHB8vx

Show HN: I made a price aggregator to find the best PC parts deals across eBay Started PC flipping as a hobby (buying used parts, building PCs, and selling them for a small profit). Found that focusing on used parts on eBay gives the best margins - like finding a GPU 20% below market, pairing it with other deals, and selling the complete build locally. I discovered that eBay marketplaces (.com, .de, .co.uk etc.) often have different prices for the same items! Or just completely different items. So I built an eBay price scanner for PC components. It scans listings across different Ebay markets, calculates median prices, and flags anything selling below market (including shipping costs to your location). It basically finds the best deal that is specific to your location. It currently tracks GPUs, CPUs, RAM and motherboards across 6 eBay markets, updating every 8 hours. Been using it myself for and it actually works! Let me know, what you think. Would love to hear your feedback on making it more useful. https://ift.tt/QoJMSVL January 30, 2025 at 01:35AM

Show HN: Mcp-Agent – Build effective agents with Model Context Protocol https://ift.tt/TOa09PJ

Show HN: Mcp-Agent – Build effective agents with Model Context Protocol Hey HN, I spent my xmas break building an agent framework called mcp-agent [1]( https://ift.tt/0osWkM8 ) for Model Context Protocol [2]. It makes it easy to build AI apps with MCP servers, and implements every pattern from the popular Building Effective Agents blog [3] as well as OpenAI’s Swarm [4]. I’m sharing it early to get community feedback on where to take it from here, and to ask for contributions. For those who aren’t familiar with MCP, I think of it as a standardized interface to let AI communicate with software via tool calls, resources and prompts. mcp-agent provides a higher level interface to build apps with MCP. It handles the connection management of MCP servers so you don’t have to. It also implements the Building Effective Agents patterns: - Augmented LLM (an LLM with access to one or more MCP servers) - Router, Orchestrator-Worker, Evaluator-Optimizer, and more - Swarm The key design principles are composability and reusability – every pattern is an AugmentedLLM itself, so you can chain them into more complex workflows. Some background: I worked on LSP [5] and language servers at Microsoft, and saw firsthand how standards and protocols can revolutionize developer workflows. Before LSP every IDE had its own esoteric ways of providing language services. LSP changed all that, and arguably made every language server better, since they can focus on improving a single implementation for all clients. I think AI development is in a similar pre-LSP space right now. There are tons of frameworks [6], every model provider has its own way of handling messages, tool calls, streaming, etc. I really think we need a protocol to standardize these patterns. Pretty soon every service is going to expose an MCP interface, and mcp-agent is about letting developers orchestrate these services into applications (i.e. build “MCP apps”). This can cover any use of an AI model that needs to interact with the world around it: - RAG pipelines and Q&A chatbots - Process automation via AI workflows/async tasks - Multi-agent orchestration, with human in the loop The repo contains examples [7] to build RAG agents, streamlit apps and more. There’s a lot left to build, like streaming support, server auth and tighter integration with MCP clients. But I wanted to share early in the hopes that you can guide me: - If you find this useful, please let me know. If it’s useful to you, I will dedicate all my time to improving it. - I really welcome contributions. If you want to collaborate, please reach out on github to help take this forward. I want to help standardize AI development, so developers a few years from now can look back with horror at the pre-MCP days. [1] - https://ift.tt/0osWkM8 [2] - https://ift.tt/MISA9wP [3] - https://ift.tt/lEx5evd [4] - https://ift.tt/oHq9Qrs [5] - https://ift.tt/09B2XRo [6] - https://xkcd.com/927/ (I understand the irony) [7] - https://ift.tt/85X3GNr https://ift.tt/0osWkM8 January 29, 2025 at 09:56PM

Show HN: I built a SaaS thanks to my wife https://ift.tt/FhgavKc

Show HN: I built a SaaS thanks to my wife I’m Michał, and I’d like to share with you the journey I went through with my wife and how, thanks to her, we built our first SaaS, PDFBolt ( https://pdfbolt.com ). I’ve been a developer for over 10 years. In 2020, I decided to build a side project to learn all aspects of app development—deployment, authentication, payments, frontend, landing pages, etc. While looking for project ideas, I came across the Indie Hackers community, where I found a simple HTML-to-PDF API project. The creator mentioned a lot of interest in it and that it was generating revenue. I thought I’d build something similar myself and learn a lot in the process. But it wasn’t easy at all. After working from 9 to 5, it’s hard to spend another few hours in front of the computer in the evening. What about other responsibilities? Groceries, cooking, cleaning, hobbies, spending time with my wife? Still, I tried, very slowly. I had breaks lasting several months, and at one point, due to mental health issues, I practically stopped working on the project altogether. My wife worked as a physiotherapist but, due to difficulties in her job, decided to switch to IT with my help, starting as a manual tester. She did it very quickly (maybe six months) and immediately found a job. In mid-2024, she started asking about my old project and insisted that we finish it. Thanks to her enthusiasm, we managed to do it very quickly. I focused on the backend, and she, in addition to testing, handled the entire frontend and landing page. Around the same time, we also adopted a dog from a shelter, which added a lot of positive energy to our lives and helped us stay motivated. In early January 2025, we officially launched the project. It’s been a long journey, and we don’t have any customers yet—we don’t even know if we will, as we have no idea about marketing :) But we’ve learned a lot and are already happy with the journey itself. As for the technical aspects, the app uses: Backend: Kotlin, Spring Boot, Postgres, Redis Frontend: React, Next.js, Docusaurus Auth: Firebase Hosting: Render (the app is Dockerized) Cloudflare R2 for file storage PDFs are generated using Chromium via Playwright. If you have any questions about the tech stack or anything else, feel free to ask! I’ll be happy to answer. Any feedback or criticism will be greatly appreciated. Thank you! :) https://pdfbolt.com/ January 30, 2025 at 12:54AM

Show HN: Cdlog: nicer directory navigation for Bash https://ift.tt/5utp6Aa

Show HN: Cdlog: nicer directory navigation for Bash https://ift.tt/PRH5vpi January 29, 2025 at 12:14AM