SOFTWARE

Prism tracks AI emissions, Super Productivity keeps tasks local, and TAMOSS brings the BBC's media API to Kubernetes

As the mercury climbs to disgusting heights, The Reg FOSS desk has picked out a few FOSS highlights from its overflowing mailbox, in case you fancy some Super Productivity while monitoring your AI habit.

Prism Carbon Tracker

The Prism carbon tracker is a neat idea for helping to instill a tiny bit of self control in the more dedicated botlickers on your development team.

It's a plugin that integrates into Visual Studio Code. As the developer uses various "online coding assistants," Prism shows an estimate of how much CO₂ they're releasing in the process of outsourcing their mental activity to a datacenter somewhere.

As the plugin's homepage says: "It captures token usage from GitHub Copilot Chat, Claude Code, and runtime LLM API calls, then calculates CO₂ equivalents and surfaces them in the sidebar, status bar, and a live dashboard."

Prism comes out of a collaborative effort between the University of Bristol and ustwo, a majority employee-owned "global digital product studio." Ustwo's Nick Hegarty said: "Developers are not the problem, but a fundamental part of the solution. By making AI emissions visible during development, we hope to create greater awareness and support better decision-making. We believe even directional estimates can be valuable if they help start conversations and encourage more thoughtful AI usage."

The emissions estimate uses "a transparent token-to-energy-to-carbon methodology informed by Green Software Foundation guidance and recent academic research."

Super Productivity

Maintainer Johannes Milan wrote to tell us about "Super Productivity." Normally a name like that would set our mental alarm bells ringing, but he did it just right. As a handy lesson for all the other breathlessly excited organizations out there, this is how to do it: with no hyperbole, and a terse text-only email with no formatting.

He told us: "I maintain Super Productivity, a mature MIT-licensed local-first task manager and time tracker. It runs on Linux, Windows, macOS, web, and mobile; has Linux packages for Flatpak, Snap, AppImage, deb, and rpm; and works without accounts or telemetry."

This is a good intro, so what does it do? It is "a local-first productivity app that is not another cloud suite: tasks, timeboxing, Pomodoro, time tracking, review, and issue tracker integrations in one offline-capable app."

The phrase "local-first" twice makes us think he has a better idea of our interests than the PR people who tell us that they liked a random article from three years ago and so should write about their new AI-based product.

If you want to see the source, it's on GitHub, and if you are more used to a bit of hyperbole, the product's homepage

TAMOSS

TAMOSS is a far more specialized tool, but may be very valuable to the people who need it. The name is short for Time-Addressable Media Open Source Store, and the product's homepage says "TAMOSS is an open source, Kubernetes-native implementation of the BBC TAMS API."

Which of course merely left us wondering what the BBC TAMS API was, but happily, the Beeb has a bit over 100 years of experience at explaining stuff. It has some pretty good educational resources of its own. We suggest starting with the BBC R&D division's TAMS in 2025 – we found the introductory video there packed a decent summary into three and a half minutes.

TAMS, short for Time-Addressable Media Store, is an API for a server that "can be used to store, query and access segmented media over HTTP." Most audio-visual media online is in the form of single, big, fat media files, which contain video, soundtrack, subtitles and more, all mixed together. This is very unhelpful if you want to edit the material, split off a soundtrack, or slow part of it down or whatever. It's a lowest common denominator sort of thing, but that's what all the tools understand and what existing servers know how to send.

The TAMS spec lets a client request and receive separate parts of this (the sound, or the video, or both, or other data) by timeline using AMWA NMOS-compatible UUIDs over HTTP. NMOS here is the Networked Media Open Specifications from the Advanced Media Workflow Association.

Did you know the BBC has its own GitHub? Us neither, but on it you'll find the open source TAMS API spec. The TAMS standard has its own site as well, and the Beeb hosts the BBC R&D white paper that the spec is based on.

TAMS is a spec, not a product. There's nothing you can deploy there, which is where LiveWyer, a London-based Kubernetes and cloud-native consultancy, came in. It developed TAMOSS as open source, and it's publishing it for free under the Apache 2.0 license. The company told us "it lets any organisation run a fully functional, TAMS-compatible media store on its own infrastructure, whether that's from a laptop or a production cluster. No proprietary licensing, no vendor lock-in." ®