Firehose

December 2025

Public projects created in December 2025, day by day: GitHub repositories, datasets and AI models.

27days · Month wrap-up

31 December 2025

Agent tooling and plugins covered a local MCP server exposing specialised tools, an agent for recommending LLMs, a work-in-progress CLI hook, a website makeover agent, shared AI chat research, and a prompt for generating documentation.

GitHub Repositories: 6

30 December 2025

An MCP server for managing BTRFS snapshots through Snapper and monitoring filesystem health.

GitHub Repositories: 1

29 December 2025

Agent tooling and prompt evaluation dominate: a consolidated list of reusable Claude development agents, tests of how AI tools handle stack request prompts, and a plan for agentic coding.

GitHub Repositories: 3

28 December 2025

A set of printable cable reference wall charts made with Nano Banana Pro.

GitHub Repositories: 1

27 December 2025

Three tools for agent and desktop workflows: a recovery boot target for using an agentic CLI on Ubuntu, a slash command that pushes agents to reuse existing solutions, and a KDE Plasma action for quick background removal.

GitHub Repositories: 3

26 December 2025

Two repositories cover desktop maintenance tooling and agentic web design. One holds planning notes for a desktop maintenance utility, the other documents a web design crew that produces offbeat designs.

GitHub Repositories: 2

25 December 2025

Notes for an AI technical documentation stack are held in a repository.

GitHub Repositories: 1

24 December 2025

A multiagent simulation of reactions to a social policy initiative and ideation notes for internal apps planning make up the new repositories, covering policy modelling and internal tool planning.

GitHub Repositories: 2

23 December 2025

A Claude app optimiser with a system prompt, slash command and agent.

GitHub Repositories: 1

21 December 2025

Two documentation repositories cover AI model behaviour: a prompt that asks Claude to identify itself and its variant at the start of each conversation, and a test of how sample intonation affects single shot voice cloning.

GitHub Repositories: 2

20 December 2025

Two home organisation repositories: a GUI and CLI for printing QR code labels on Brother QL-700 printers, and a taxonomy for a Homebox fork.

GitHub Repositories: 2

18 December 2025

Fine-tuning material and prompt notes for cleaning up transcribed text, plus research on voice control platforms that run on Linux. A curated audio dataset for training speech-to-text cleanup models rounds out the repositories.

GitHub Repositories: 3 · Datasets: 1

16 December 2025

Four repositories cover AI prompting and tooling: prompts for audio transcript formatting, a slash command for rewriting technical documentation, a macropad layout for dictation, and an MCP server for comparing AI model pricing.

GitHub Repositories: 4

15 December 2025

Project and app idea pipelines, a long-form audio preparation tool, and a list of vendor CLIs with agentic capabilities covered ideation and voice tooling. Two MCP servers handled a manually managed LAN directory and Wasabi S3 operations, alongside a subreddit collection, a vibe-coding graphic, and an AI experiment ideator.

GitHub Repositories: 8

14 December 2025

A repository documents planning for an AI model manager.

GitHub Repositories: 1

13 December 2025

Two repositories cover career planning for programmers and a local MCP handler for secrets that pairs with a cloud API.

GitHub Repositories: 2

12 December 2025

A voice note tool that transcribes and tidies recordings with cloud audio models sits alongside an evaluation of how well major text-to-image models render Hebrew text.

GitHub Repositories: 1 · Datasets: 1

11 December 2025

Home monitoring setup notes, an evaluation of text to image models on Hebrew text, and a set of shop recommendations mostly in Israel make up the repositories.

GitHub Repositories: 3

10 December 2025

Speech recognition evaluations covered an index of ASR benchmarks, a transcription cleanup comparison across cloud audio models, and a test of Whisper fine-tunes against stock Whisper on local inference. The accompanying datasets were a small speech-to-text set of 92 audio samples with ground truth transcriptions and a fine-tune evaluation set.

GitHub Repositories: 3 · Datasets: 2

9 December 2025

A tool for managing starred items on GitHub and documentation for using Claude Code with MCPM cover developer tooling. Speech recognition work includes an evaluation of how speaking pace and background noise affect transcription accuracy, plus the annotated audio dataset behind it.

GitHub Repositories: 3 · Datasets: 1

8 December 2025

Audio multimodal testing prompts, a karaoke video generation script, and documentation covering bug reports and component compatibility make up the new repositories.

GitHub Repositories: 4

7 December 2025

Audio-focused multimodal AI material covers model profiles, benchmarks and documentation for speech recognition and transcription. Two smaller tools provide a utility for populating Claude workspaces and a snapshot of cloud speech-to-text pricing.

GitHub Repositories: 3

6 December 2025

GPU compatibility notes for an AMD Radeon 7700 sit alongside documentation on a speech-to-text processing chain and microphone test samples for transcription accuracy, plus a written response to a user interview.

GitHub Repositories: 4

5 December 2025

A Claude Code plugin for planning and burning batch optical disc archives on Ubuntu, with copy multipliers for offsite duplicates, and a prototype chat interface for lashon hara questions and answers.

GitHub Repositories: 2

3 December 2025

Repositories cover agentic AI reporting and local infrastructure: an MCP example for long-form report generation, an MCP server for saving files to a NAS over NFS, and documentation for running local inference on AMD GPUs. Other work covers smart home Halacha, prompt experiments comparing speech cleanup approaches, and a dataset of captioned Israel photographs.

GitHub Repositories: 5 · Datasets: 1

2 December 2025

A whiteboard cleanup pipeline and a work in progress server for cloud speech recognition models cover image tidying and transcription. A multiagent geopolitical simulation exploring problem solving and bridge building in the Middle East rounds out the repositories.

GitHub Repositories: 3

1 December 2025

Configurations for AI agent virtual panels explore simulated voting bodies and thinking groups, with sets of panels for testing a frontier application of AI personas.

GitHub Repositories: 1