Hi everybody, I recently rewrote my old translation workflow and decide to finally share it: The title says it all; the idea of WordWeave is to make translation effortless while providing enough customisation to make the workflow fit you perfectly. Two examples for how WordWeave could look like for the same translation:Some special features of this workflow:Support for local and online translation: Local translation uses a lean implementation of Argos (~200ms per translation), online uses Google Translate. Each translator has its own keyword, you can use either or both. The Google Translate module uses an unofficial public API so it needs no API key, but it may break at any time.Each input produces a ranked list of candidate translation, not just a single "best guess". A user-configurable parameter lets you dial in how many candidates are kept (trade-off between accuracy & fuzziness).(the initial reason I wanted to build WordWeave) Ping-pong translation: translate a word, press Tab, translate the result back, repeat. A word in one language often maps to several in another, so bouncing back and forth works like a cross-language thesaurus.The translation pair (source → destination) must be specified for each translation. Languages are specified either by two-letter ISO 639-1 codes, or by QuickCodes, single-letter abbreviations chosen by the user (e.g., as in the demo images above, g → German, e → English). No delimiter is placed between the two languages, therefore, for each query, you must choose between either ISO 639-1 or QuickCode.A national flag - or any other emoji - can be chosen by the user to represent each language (e.g., see above, English → 🇺🇸 or 🌍️, ...)For the local Argos module: models are downloaded from the official Argos index via a separate keyword, and are one-way only (English → German and German → English are two separate models). Almost all Argos translation models are either x → English or English → x. When translating between two languages that don't have a direct model path (e.g., German → Spanish), WordWeave's router automatically tries to find a path between the two languages (pivot translation). The pivot is usually English (e.g., German → English → Spanish). This step works well but is not lossless. Expect slightly worse results and slightly longer translation time.This workflow requires the uv package manager, and WordWeave downloads dependencies on first use. I know this disqualifies it from the Gallery, but I didn't want to deal with shipping binaries (and having to notarise them) so this is where the project is currently at - if you have a different suggestion, feel free to contribute via GitHub. Thanks! :) pyproject.toml lists exactly which packages are downloaded (excl. transitive dependencies - but if you worry about this kind of stuff, you likely already know how to check for those).Anyway, if this sounds like something right up your alley, check out the repo to for some more info:https://github.com/paraversal/WordWeave/and if you're ready, download the ultimate translation workflow at:https://github.com/paraversal/WordWeave/releases/latest
| # | Наименование новости | Тональность | Информативность | Дата публикации |
|---|---|---|---|---|
| 1 | Folo RSS Reader — Read timeline with Alfred | 0 | 8.57 | 05-10-2026 |
| 2 | Новости науки свайпами: serverless на AWS, LLM за цент на статью и 12 тестировщиков для Google Play | 0 | 8.77 | 27-09-2026 |
| 3 | GPT, Claude, and Gemini in one browser tab for $39.99, for life | 0 | 14.36 | 03-10-2026 |
| 4 | Copy Fix Prompt, diagnose and fix failures with any AI coding agent | 0 | 6.48 | 12-08-2026 |
| 5 | Fix failing workflows with Claude Code from the CircleCI UI | 0 | 7.13 | 06-08-2026 |
| 6 | The New SDLC From Google Has a Harness. It Doesn't Have a Team. | 0 | 14.4 | 30-09-2026 |
| 7 | EIP Editing Office Hour (EIP + ERC ) Meeting #113, Sep 15, 2026 | 0 | 20.46 | 07-09-2026 |
| 8 | All Core Devs - Testing (ACDT) #96, September 14, 2026 | 0 | 16.55 | 07-09-2026 |
| 9 | Fix failing workflows with Claude Code, directly from the CircleCI UI | 0 | 15.94 | 06-08-2026 |
| 10 | Gemini 4 arrasa en los benchmarks y vuelve a meter a Google en la carrera de la IA | 0 | 5.98 | 01-10-2026 |