Show HN Daily Selection (2025-07-01)

• TokenDagger – A tokenizer faster than OpenAI's Tiktoken
• New Ensō – first public beta
• We're two coffee nerds who built an AI app to track beans and recipes
• A continuation of IRS Direct File that can be self-hosted
• C.O.R.E – Opensource, user owned, shareable memory for Claude, Cursor
• Private real-time dictation app for Mac
• Audiopipe – Pipeline for audio diarization, denoising and transcription
• Attach Gateway – one-command OIDC/DID auth for local LLMs
• QuizKnit, an open source quiz creator
• MailMap – Turn emails into interactive stories on Google Maps
• Semantic-dictionary – A Python dictionary with semantic lookup
• DotnetEbpf - Write Linux eBPF kernel applications in C#
• Ravana, Multi AI Assistant Browser Desktop App
• My Cross-Platform MySQL Parser
• AI image alt text generation tool – Turn images into text with AI
• Sysadmin.ca Free Tools and Real-World Resources for Everyday Sysadmins
• Cheesy Mamas: Local only code editor with Git and Bash support

TokenDagger – A tokenizer faster than OpenAI's Tiktoken [Developer Tools]

239 points, 66 comments

TokenDagger is a high-performance, drop-in replacement for OpenAI's Tiktoken tokenizer, written in C++ 17 with Python bindings, offering significant speed improvements.

• Written in C++ 17 with thin Python bindings for compatibility
• Maintains exact BPE vocab and special-token rules as Tiktoken
• 4x faster single-threaded tokenization
• 2-3x higher throughput on large text files (1GB tested)
• Uses a faster jit-compiled regex engine and simplified algorithm for speed gains
• Includes benchmarking code for performance validation

Discussion Link, Original Link

New Ensō – first public beta [Productivity]

210 points, 80 comments

Ensō is a writing tool designed to help users achieve a flow state by separating writing from editing.

• Focuses on uninterrupted writing by disabling editing during sessions
• Built based on 6 years of experience and 2 million words of daily writing
• Simplified and more accessible design in its first public beta
• Encourages creative flow by minimizing self-editing distractions

Discussion Link, Original Link

We're two coffee nerds who built an AI app to track beans and recipes [Life Assistant]

34 points, 20 comments

An AI-powered iOS app for coffee enthusiasts to track beans, recipes, and brewing details with automated data extraction and organization.

• Auto-detects coffee bag details (roaster, origin, process, etc.) via photo snap
• Extracts structured recipes from YouTube links or photos
• Logs grind size, roast timeline, ratings, and notes in a clean UI
• Provides coffee year-in-review analytics for tracking habits and favorites
• Includes AI assistant for brew-related queries and bean facts

Discussion Link, Original Link

A continuation of IRS Direct File that can be self-hosted [Finance]

4 points, 0 comments

A self-hostable continuation of the IRS's open-sourced Direct File tax tool, updated for the next tax season.

• Based on the IRS's open-sourced Direct File project
• Self-hostable for independent use
• Updated to be ready for the next tax season
• Documentation available online for reference

Discussion Link, Original Link

C.O.R.E – Opensource, user owned, shareable memory for Claude, Cursor [Developer Tools]

3 points, 0 comments

C.O.R.E is an open-source, user-owned knowledge graph that enables multiple LLMs (like ChatGPT, Claude, and Cursor) to share and query persistent context.

• Portable context: Avoids vendor lock-in by allowing cross-tool memory sharing.
• Temporal + relational: Stores full version history of facts with timestamps and provenance.
• User-controlled: Choose between local (Docker) or hosted deployment, with control over private vs. synced memories.
• Traceable answers: Hover to see who/when/why a fact was added or modified.
• Free hosted tier available for testing.

Discussion Link, Original Link

Private real-time dictation app for Mac [Audio]

9 points, 2 comments

Private real-time dictation app for Mac.

• Real-time dictation
• Privacy-focused
• Mac-only application

Discussion Link, Original Link

Audiopipe – Pipeline for audio diarization, denoising and transcription [Audio]

4 points, 0 comments

Audiopipe is a streamlined tool for denoising, diarization, and transcription of audio using demucs, pyannote, and insanely-fast-whisper.

• Combines denoising, diarization, and transcription in a single command
• Designed for transcribing podcasts and Dungeons and Dragons sessions
• Includes a WASM-based UI for loading transcriptions and audio
• Open for contributions and feedback

Discussion Link, Original Link

Attach Gateway – one-command OIDC/DID auth for local LLMs [Developer Tools]

4 points, 0 comments

Attach Gateway is a one-command OIDC/DID authentication solution for local LLMs, simplifying auth setup for model servers like Ollama or vLLM.

• Verifies OIDC/DID JWTs
• Adds X-Attach-User and X-Attach-Session headers for shared identity
• Includes optional /a2a/tasks/send endpoint for A2A hand-offs
• Mirrors prompts and completions to Weaviate
• Runs in Docker with quick setup via pip install attach-dev

Discussion Link, Original Link

QuizKnit, an open source quiz creator [Education]

3 points, 0 comments

QuizKnit is an open-source quiz creator that generates quizzes from text or historical categories.

• Creates quizzes by pasting text or selecting from historical categories
• Allows users to take, save, and share quizzes via URL
• Started as an OpenAI API experiment, now open-sourced
• Includes working demo at quizknit.com
• Frontend and backend code available on GitHub

Discussion Link, Original Link

MailMap – Turn emails into interactive stories on Google Maps [Productivity]

3 points, 0 comments

MailMap turns emails with geotagged photos into interactive map stories instantly.

• No app downloads or registrations required for basic use
• Automatically extracts GPS coordinates from image EXIF data
• Real-time processing and map updates
• Supports JPEG, PNG, and GIF formats
• Community moderation and story voting
• Mobile-responsive Google Maps integration
• Direct map editing for registered users

Discussion Link, Original Link

Semantic-dictionary – A Python dictionary with semantic lookup [Developer Tools]

4 points, 0 comments

A Python dictionary that uses semantic similarity for key lookup, enabling flexible matching of semantically related terms.

• Uses embedding models to calculate semantic similarity between keys
• Supports flexible lookup for synonyms or related terms
• Useful for AI applications handling varied or evolving key formats
• Available as a distributable Python package

Discussion Link, Original Link

DotnetEbpf - Write Linux eBPF kernel applications in C# [Developer Tools]

4 points, 0 comments

A .NET to C transpiler enabling Linux eBPF kernel applications to be written in C#.

• Transpiles C# code to C for embedded development
• Supports writing Linux eBPF kernel applications in C#
• Provides better syntax than C and unit testing capabilities
• Future goal: seamless integration between C# kernel and user-space eBPF applications

Discussion Link, Original Link

Ravana, Multi AI Assistant Browser Desktop App [Life Assistant]

3 points, 1 comments

Multi AI assistant browser desktop app.

• Supports multiple AI assistants
• Designed for browser and desktop use
• Combines functionality in a single app

Discussion Link, Original Link

My Cross-Platform MySQL Parser [Developer Tools]

4 points, 0 comments

A cross-platform MySQL parser extracted from TiDB, optimized for WASM compatibility and independent usage.

• Extracted from TiDB's parser to avoid dependency on TiDB's monolithic structure
• Fully compatible with MySQL syntax
• Designed for cross-platform use, including WASM support
• Includes an interactive demo for testing queries
• Independent Go package for easier integration

Discussion Link, Original Link

AI image alt text generation tool – Turn images into text with AI [Image]

4 points, 0 comments

AI-powered tool and WordPress plugin for automatic alt text generation from images.

• Generates alt text for images automatically using AI
• Available as an online tool and WordPress plugin
• Paid service with free credits for testing
• Designed by a freelance WordPress developer

Discussion Link, Original Link

Sysadmin.ca Free Tools and Real-World Resources for Everyday Sysadmins [Productivity]

4 points, 0 comments

A free, no-signup resource hub for sysadmins offering practical tools and templates.

• Built by an IT admin for IT admins
• Privacy-respecting with no fluff
• Includes cheat sheets, outage trackers, and vendor call prep templates
• Completely free with only one ad
• Focuses on real-world utility over theory

Discussion Link, Original Link

Cheesy Mamas: Local only code editor with Git and Bash support [Developer Tools]

4 points, 2 comments

A local-first, Git-integrated code editor for Linux with built-in Bash support and no telemetry.

• Local-first design with no telemetry, sync, or accounts
• Multi-tab editor with syntax highlighting for Python, C, and LaTeX
• Built-in execution for Python, C (via gcc), and LaTeX (via pdflatex)
• Embedded Git history sidebar with per-file commit tracking
• Visual diff and granular revert options for individual lines
• Dark UI with persistent tab state and no background processes
• Single-instance handling with file path relay for new tabs
• Optional default file handler for .py, .c, .tex, and .sh files
• Fully offline, installable via shell script with .desktop integration

Discussion Link, Original Link