15:[[47669920,{"by":"jaafarskafi","descendants":0,"id":47669920,"kids":[47719175,47672786],"score":1,"text":"$16","time":1775527766,"title":"Show HN: MCP plugin that lets Claude autonomously pay for APIs via Lightning","type":"story","url":"https://github.com/jeffskafi/OpenBolt"}],[47671303,{"by":"krish678","descendants":0,"id":47671303,"kids":[47672783],"score":3,"text":"I’ve been working on a small system to understand how applications can stay up even when backends fail.

The idea is simple:\ninstead of sending requests to a single backend, route them through a layer that can switch to another backend if something goes wrong.

It:

checks backend health (latency, errors)\navoids unhealthy servers\nretries requests on another backend if needed

It’s designed as:

a fast routing layer (Rust)\na simple control API (Python)\nshared state via Redis

One thing I found interesting is that failover only works before the response starts — after that, switching isn’t possible.

Still early and mostly an experiment to understand failover and reliability better. This begun as internal experiment, after recent region outages.

Curious how others approach this problem in production systems.","time":1775542349,"title":"Show HN: System that keeps apps running when a server or region goes down","type":"story"}],[47672321,{"by":"mesahm","descendants":5,"id":47672321,"kids":[47672341,47672780,47673486,47672558],"score":3,"time":1775551343,"title":"Show HN: Niquests – A Requests fork, three years of catching Python's HTTP up","type":"story","url":"https://github.com/jawah/niquests/tree/v3.18.4"}],[47657165,{"by":"cloudkj","descendants":1,"id":47657165,"kids":[47658799],"score":1,"text":"I was looking through some of my past personal projects tinkering with reinforcement learning, and noticed that the REST/HTTP API for the OpenAI gym available at the time is no longer supported. The API was pretty useful back then since most of ML and deep learning hadn't quite stabilized on the various Python libraries today.

gymnasium-http-api is an attempt at bringing back the same type of language-agnostic support. It mostly wraps the forked and supported Gymnasium library, with some endpoints specifically for making it easier to render and visualize the training and learning process.

Check out the README for some examples. Hope others find it useful!","time":1775451540,"title":"Show HN: REST API for Gymnasium (fka OpenAI Gym) reinforcement learning library","type":"story","url":"https://github.com/cloudkj/gymnasium-http-api"}],[47658617,{"by":"anton_salcher","descendants":13,"id":47658617,"kids":[47658796,47658675,47658667,47662373,47664226,47667937,47661520],"score":20,"text":"Hey HN,

I built Pace, a Claude Connector that lets you connect all your wearables (Garmin, Polar, Whoop, 20+) with Claude.

You connect your devices once and can analyze your data with Claude. No Dashboard needed, just in natural language.\nI already use it everyday, especially the visualization tool in Claude makes this really cool to use.

Tools include: overview, sleep, training, activity, samples and trends.

Tech Stack: Python (FastMCP), Google Cloud Run, Google Cloud SQL (PostgreSQL) and Firebase

It is live and free to try (no Claude Pro/MAX Plan needed)..

Would love feedback, especially from people who've built MCP servers or use wearables seriously.","time":1775467365,"title":"Show HN: I just built a MCP Server that connects Claude to all your wearables","type":"story","url":"https://pacetraining.co/"}],[47650438,{"by":"srodriguezp","descendants":0,"id":47650438,"kids":[47658787],"score":5,"text":"A prompt tuned for Llama 3 often degrades on Mistral or Qwen — same task, different behavioral surface. Identa automates the recalibration.\nIt implements two things from the PromptBridge paper (arXiv:2512.01420):

A transfer engine that learns a mapping between model behaviors using source/target prompt pairs\nA MAP-RPE evolutionary loop that iteratively improves candidates against a scoring function until behavioral parity is reached

Works fully local via Ollama. Also supports OpenRouter for cross-hosted runs. No telemetry, no cloud dependency.\nBuilt with Python, Typer, Pydantic.\nHappy to go deep on the calibration algorithm or the tradeoffs in the scoring design.

https://github.com/shepax/identa-agent","time":1775403017,"title":"Show HN: Identa – CLI to calibrate prompts across local LLMs","type":"story"}],[47613359,{"by":"ryaker","descendants":1,"id":47613359,"kids":[47624725],"score":2,"text":"$17","time":1775131885,"title":"Show HN: SparrowDB, SQLite of graph DB, written in Rust, Cypher compatible","type":"story","url":"https://github.com/ryaker/SparrowDB"}],[47614404,{"by":"CodeReclaimers","descendants":1,"id":47614404,"kids":[47618874,47624723],"score":5,"text":"$18","time":1775137511,"title":"Show HN: Symbolic regression as an MCP tool (SINDy and PySR, free, no install)","type":"story","url":"https://occam.fit/"}],[47620103,{"by":"mennylevinski","descendants":2,"id":47620103,"kids":[47624720,47622321],"score":2,"text":"Secuditor Lite is a free, Python-based diagnostic security tool for Windows endpoints and networks with a friendly GUI.

It helps identify system vulnerabilities, improve device Operational Security (OPSEC), detect network elements, run internet speed test, and generate structured audit reports, all in one place.","time":1775163554,"title":"Show HN: I built a Python-based app for Windows security diagnostics","type":"story","url":"https://github.com/secuditor/secuditor-lite/tree/main"}],[47601570,{"by":"bitscraper","descendants":1,"id":47601570,"kids":[47612936,47601571],"score":12,"time":1775054270,"title":"Csp-toolkit – Python library to parse, analyze, and find bypasses in CSP headers","type":"story","url":"https://chs.us/2026/03/csp-toolkit/"}],[47612046,{"by":"mesahm","descendants":1,"id":47612046,"kids":[47612110,47612903],"score":1,"time":1775122257,"title":"Show HN: A drop-in replacement and memory-safe TLS back end for Python","type":"story","url":"https://github.com/jawah/rtls"}],[47611994,{"by":"kdeldycke","descendants":2,"id":47611994,"kids":[47612892,47612207],"score":9,"text":"$19","time":1775121849,"title":"Show HN: Extra-Platforms, Python library to detect OS, arch, shell, CI, AI","type":"story","url":"https://github.com/kdeldycke/extra-platforms"}],[47588818,{"by":"unfavalen","descendants":0,"id":47588818,"kids":[47588901,47598384],"score":1,"text":"I kept running into the same issue with AI: every conversation starts from zero, even when I already solved the same problem before. I still had to repeat constraints, preferences, and workflow every time. So I built AEP (Agent Experience Protocol), a simple way to save successful AI workflows such as intent, constraints, preferences, steps, failure traps, and success checks, and reuse them later so future tasks start aligned instead of from scratch. It lives as repo-local JSON in .agent/aep/, is agent-agnostic, and works more like “Git for how you work with AI” than prompts or memory. Curious if others are hitting the same problem and whether something like this would be useful in tools like Cursor, ChatGPT, or Claude.","time":1774970943,"title":"Show HN: I built a way for AI to remember how you like things done","type":"story","url":"https://github.com/Robi-Labs/AEP"}],[47589764,{"by":"asieltechlab","descendants":1,"id":47589764,"kids":[47598366,47589765],"score":1,"time":1774974227,"title":"AI agent that writes and hot-reloads its own Python modules at runtime","type":"story","url":"https://github.com/asieltechlab/asiel-core"}],[47542570,{"by":"nishalk","descendants":3,"id":47542570,"kids":[47598341,47547862,47619254,47558353],"score":3,"text":"$1a","time":1774619018,"title":"Show HN: SpedFind – An open-source internet speed test (React/Python and CLI)","type":"story"}]]

user:	thitami
created:	March 25, 2026
karma:	2