---
name: Radar Daily Briefings
description: A personalized daily briefing for WordPress and software engineers.
date: 2026-11-16
--- 

--------------------------------------------------------------------------------

---
story_id: story_983f260c1af84d928de0ad270e56e830
authors: []
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: hugging-face-blog
section: ai
tags:
  - xybench
  - llm-evaluation
  - benchmark-dataset
  - misconception-identification
  - intent-understanding
title: Ai2 releases XYBench benchmark for mistaken user assumptions
url: https://huggingface.co/blog/allenai/xybench
why_read: XYBench separates literal compliance from goal-directed assistance, giving engineers metrics for evaluating intent handling in prompts and model training.
status: released
source_published_at: 2026-11-13T17:02:35.000Z
source_external_id: https://huggingface.co/blog/allenai/xybench
source_adapter: rss
interest_score: 8
utility_score: 7
novelty_score: 8
depth_score: 9
impact_score: 7
---

Ai2 has released XYBench, a benchmark of 8,115 questions designed to test whether AI assistants recognize mistaken assumptions and address users’ underlying goals. The dataset covers programming and everyday activities, including cooking and home repair.

XYBench measures whether responses solve the stated request or inferred goal, which aspect they emphasize, and whether they discourage the mistaken approach. The article reports that five evaluated models more often addressed stated requests than underlying goals and primarily focused on the underlying goal in only 35–41% of Stack Overflow responses.

The benchmark also reports a gap between model-written answers and human references on misconception identification. It currently covers English-language, mainly procedural queries.

--------------------------------------------------------------------------------

---
story_id: story_77fec7ee93674a639d6b1b4482260d7f
authors: []
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: hugging-face-blog
section: ai
tags:
  - decision-models
  - structured-prediction
  - probabilistic-classification
  - zero-shot-classification
  - transformers
  - typed-decisions
title: Hugging Face presents a typed interface for decision models
url: https://huggingface.co/blog/decision
why_read: Engineers can compare decision-model interfaces with classification and structured generation when designing probabilistic application control flows.
status: released
source_published_at: 2026-11-10T00:00:00.000Z
source_external_id: https://huggingface.co/blog/decision
source_adapter: rss
interest_score: 8
utility_score: 8
novelty_score: 7
depth_score: 8
impact_score: 7
---

Hugging Face presents decision models as a way to evaluate application-defined questions and return typed decisions with probabilities, rather than generating a complete text response. The article documents the interface and associated Transformers support.

A request contains state and questions. Choice selects among named alternatives, Score computes the expected value of an ordered distribution, and Noul returns the probability of a binary statement. The article also describes DecisionPipeline, AutoModelForDecision, DecisionConfig, and a /v1/systemone endpoint.

The approach can make routing and structured application decisions explicit. However, Hugging Face characterizes the area as young and fast-moving, and the supplied material provides no independent performance or calibration comparison.

--------------------------------------------------------------------------------

---
story_id: story_df3284cd8c334a3d9e401faa2c3604ab
authors:
  - yorickpeterse.com by yorickpeterse
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: lobsters
section: open-source
tags:
  - codeberg-migration
  - repository-migration
  - forgejo-actions
  - ci-runners
  - self-hosting
  - storage-quotas
title: Maintainer documents migration of 70+ repositories to Codeberg
url: https://yorickpeterse.com/articles/migrating-from-github-to-codeberg
why_read: The migration details expose concrete operational tradeoffs for maintainers evaluating Codeberg and Forgejo as GitHub alternatives.
status: released
source_published_at: 2026-11-16T05:10:34.000Z
source_external_id: https://lobste.rs/s/rhoopo
source_adapter: rss
discussion: https://lobste.rs/s/rhoopo/migrating_from_github_codeberg
discussions:
  - source: lobsters
    url: https://lobste.rs/s/rhoopo/migrating_from_github_codeberg
interest_score: 8
utility_score: 8
novelty_score: 7
depth_score: 8
impact_score: 6
---

Yorick Peterse documents migrating actively maintained personal and Inko repositories from GitHub to Codeberg, covering more than 70 repositories. The migration also replaced GitHub Actions with Forgejo Actions and introduced self-hosted runners.

The account describes importer limitations, including missing user mapping and unsupported GitHub issue types. Codeberg's shared runners were considered too constrained for the workload, so Peterse used an M1 Mac Mini and AMD64 hardware for multi-platform and multi-architecture testing. Buildah was used for container images because nested Podman or Docker builds were unavailable on shared runners.

Peterse reports slower service performance, recent outages, and restrictive storage quotas. The Inko organisation therefore continued using GitHub's container registry for some images while awaiting a quota response.

--------------------------------------------------------------------------------

---
story_id: story_375568d757c347f08131154b66daa64d
authors:
  - iter.ca via pgr0ss
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: lobsters
section: security
tags:
  - universal-xss
  - ublock-origin
  - browser-extension-security
  - trusted-filter-injection
  - firefox-mv2
  - element-picker
title: uBlock Origin fixes website-driven universal XSS
url: https://iter.ca/post/ubo-uxss
why_read: Security engineers can understand the extension-origin attack chain and the message-authentication control used in the fix.
status: fixed
source_published_at: 2026-11-16T05:12:27.000Z
source_external_id: https://lobste.rs/s/x0ykkw
source_adapter: rss
discussion: https://lobste.rs/s/x0ykkw/universal_xss_ublock_origin_from_website
discussions:
  - source: lobsters
    url: https://lobste.rs/s/x0ykkw/universal_xss_ublock_origin_from_website
interest_score: 9
utility_score: 8
novelty_score: 8
depth_score: 9
impact_score: 8
---

A technical writeup describes a fixed universal XSS in uBlock Origin’s Firefox MV2 version. A malicious website could use the extension’s element picker to add arbitrary trusted filters, according to the author.

The attack involved discovering the extension’s internal UUID, embedding the web-accessible picker in an iframe, and sending crafted messages through a MessageChannel. A newline in a controlled option field could inject a trusted scriptlet filter. The website could then overlay or hide the iframe to induce a click.

Trusted filters could execute JavaScript in other sites’ contexts. One described fix authenticates picker messages with a secret passed by the content script; the supplied evidence gives no fixed version.

--------------------------------------------------------------------------------

---
story_id: story_8bb1e719b12a47bea0b363a50b0b4b79
authors:
  - Corey Hall
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: make-wordpress-community
section: community
tags:
  - wordpress-community
  - student-clubs
  - campus-connect
  - meetups
  - wordcamps
  - contributor-programs
title: WordPress newsletter highlights student-led learning and event formats
url: https://make.wordpress.org/community/2026/11/16/a-high-schools-first-wordpress-credits-cohort-students-teaching-students-and-one-campus-event-split-across-three-labs
why_read: Organizers can adapt the documented student-teaching, workshop, and small-group event patterns to their own WordPress programs.
status: released
source_published_at: 2026-11-16T03:13:07.000Z
source_external_id: https://make.wordpress.org/community/?p=107698
source_adapter: rss
interest_score: 5
utility_score: 6
novelty_score: 6
depth_score: 5
impact_score: 6
---

The WordPress Community’s November 2026 newsletter reports activity across WordPress Credits, student clubs, Campus Connect, meetups, and WordCamps. It says WordPress Credits had passed 50 institutions and recorded 364 graduates across 17 countries as of October 19.

The roundup describes a student-led club in Ajmer that ran four sessions, ending with a two-day portfolio build where every participant created a site. In Kolhapur, organizers divided Campus Connect #4 across three computer labs with three facilitators.

For organizers, the issue offers adaptable formats including hands-on Studio workshops, online Ask Anything sessions, and discussion-provoking talks. Operational detail and independent verification are limited.

--------------------------------------------------------------------------------

---
story_id: story_4a64cfb2535f4ea6a4f058fbc5b2653a
authors:
  - WebDecoy
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: make-wordpress-polyglots
section: wordpress
tags:
  - wordpress-localization
  - translation-review
  - ai-assisted-drafts
  - stable-readme
  - locale-maintenance
title: WebDecoy requests review of AI-assisted readme drafts
url: https://make.wordpress.org/polyglots/2026/11/15/webdecoy-readme-review-de_de-es_es-it_it
why_read: WordPress localization contributors can review the linked locale drafts and provide language corrections without seeking translation-editor rights.
status: proposed
source_published_at: 2026-11-15T19:03:29.000Z
source_external_id: https://make.wordpress.org/polyglots/?p=75768
source_adapter: rss
interest_score: 4
utility_score: 5
novelty_score: 3
depth_score: 2
impact_score: 3
---

WebDecoy has submitted 15 stable-readme suggestions for each of the de_DE, es_ES, and it_IT locales and is requesting community language review. The document points reviewers to the corresponding WordPress.org translation pages.

WebDecoy says the drafts are AI-assisted and cover the plugin’s short description, AI traffic, and local monitoring. It reports that product names, numbers, and HTML were checked, but does not claim native-speaker review.

The request is a proposed localization-review activity, not a software release or merged change. WebDecoy asks for wording corrections or guidance to the appropriate local review channel and explicitly says it is not requesting Project Translation Editor rights.

--------------------------------------------------------------------------------

---
story_id: story_a4bb9abf6b734c9b8cb0a3546a219d4c
authors:
  - harshjadav003
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: make-wordpress-polyglots
section: wordpress
tags:
  - wordpress-localization
  - hindi-translation
  - translation-editor
  - plugin-translation
title: BrightBoard Notes requests Hindi Project Translation Editor access
url: https://make.wordpress.org/polyglots/2026/11/16/hello-hindi-team-and-polyglots
why_read: WordPress contributors can understand the proposed Hindi localization workflow and the project-specific access being requested.
status: proposed
source_published_at: 2026-11-16T06:15:01.000Z
source_external_id: https://make.wordpress.org/polyglots/?p=75784
source_adapter: rss
interest_score: 3
utility_score: 4
novelty_score: 2
depth_score: 2
impact_score: 2
---

The BrightBoard Notes plugin author has proposed adding Polyglotsmed as a Project Translation Editor for the Hindi locale. The request is addressed to the WordPress Polyglots team and concerns plugin version 1.2.2.

The post reports 34 suggested strings pending review, including 16 waiting and one fuzzy string in the Stable translation project. It says Polyglotsmed contributed the Hindi suggestions and should be able to review and approve them.

This is an access request, not a completed role change. The author also says a Hindi General Translation Editor may review the strings directly, leaving the final review arrangement unresolved.

--------------------------------------------------------------------------------

---
story_id: story_99686618ff4847acafeeeea89484bf63
authors: []
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: simon-willison
section: security
tags:
  - datasette
  - access-control
  - database-download
  - security-fix
title: Datasette 0.65.6 backports a database access-control security fix
url: https://simonwillison.net/2026/Nov/15/datasette-2
why_read: Operators using custom Datasette database routes should assess exposure and upgrade to the released version.
status: released
source_published_at: 2026-11-15T23:18:17.000Z
source_external_id: https://simonwillison.net/2026/Nov/15/datasette-2/
source_adapter: atom
interest_score: 7
utility_score: 8
novelty_score: 3
depth_score: 4
impact_score: 6
---

Datasette 0.65.6 is a released update containing a minor security fix backported from Datasette 1.0a43.

The release note describes a route-based access-control issue. When a database was registered with `add_database(db, name=..., route=...)` and the route differed from its name, an actor denied access could still download it from `/<route>.db`.

Datasette operators using custom database routes should review their configuration and upgrade to 0.65.6. The supplied release note does not explain the implementation, testing, or the full range of affected versions.

--------------------------------------------------------------------------------

---
story_id: story_a5861e18cc7942849f4e25a2f3af187c
authors: []
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: simon-willison
section: web-development
tags:
  - datasette-apps
  - sandboxed-javascript
  - sql-applications
  - url-fragments
  - cross-app-linking
  - datasette-agent
title: datasette-apps 0.4a0 adds cross-app URL fragment linking
url: https://simonwillison.net/2026/Nov/15/datasette-apps
why_read: The release enables record-focused and search applications to share navigation through URL fragments and Datasette Agent integration.
status: released
source_published_at: 2026-11-15T04:02:45.000Z
source_external_id: https://simonwillison.net/2026/Nov/15/datasette-apps/
source_adapter: atom
interest_score: 6
utility_score: 7
novelty_score: 7
depth_score: 5
impact_score: 5
---

datasette-apps 0.4a0 adds URL-fragment linking for sandboxed HTML and JavaScript applications that run inside Datasette. The release describes these apps as capable of running SQL queries against Datasette data.

With the new support, an app can read and use fragments from its URL. Supported URL schemes are also reported to Datasette Agent, which can help write applications that link to one another.

The documented use case separates record search from record details: one app finds records and links to representations in another. The supplied announcement does not include implementation details, performance data, or compatibility results.

--------------------------------------------------------------------------------

---
story_id: story_73840a9394874e9ab9efb0da29429304
authors: []
date: 2026-11-16
generated_at: 2026-11-16T07:30:25.156Z
source: simon-willison
section: programming
tags:
  - datasette
  - streaming-renderers
  - response-stream
  - security-fixes
  - view-transitions
  - web-components
title: Datasette 1.0a43 adds streaming renderers and view transitions
url: https://simonwillison.net/2026/Nov/15/datasette
why_read: Datasette users and plugin authors can assess new streaming APIs, query-metadata exposure fixes, and browser-dependent navigation improvements.
status: released
source_published_at: 2026-11-15T23:20:27.000Z
source_external_id: https://simonwillison.net/2026/Nov/15/datasette/
source_adapter: atom
interest_score: 7
utility_score: 8
novelty_score: 7
depth_score: 7
impact_score: 6
---

Simon Willison has released Datasette 1.0a43, an open-source tool for exploring and publishing data. The release adds streaming output renderers, security fixes, and browser-oriented interface improvements.

Built-in TSV and CSV exports now use streaming. Plugins can register support for streaming every row through the new Response.stream() method. The row-page JSON representation was redesigned, and new copy-button and modal utilities were added.

One security fix addresses exposure of private stored-query titles and SQL without allowing execution. CSS multi-page view transitions make navigation feel more like an SPA, with support reported for Safari and Chrome; Firefox support remains prospective in the supplied note.
