Radar Daily Briefings
A clearer signal for WordPress and engineering
/\_/\
(=^.^=)
(")_(")
/\_/\
(=^.^=)
(")_(")
Revoked eMudhra certificates remain active on Indian government sites
A report published October 23 says 156 hosts in India’s government website directory were still serving certificates revoked by eMudhra on October 19. The report also identifies four MeitY-related bodies among affected sites and says certificate rotations had not yet occurred for the listed hosts.
The reported revocations are recorded in eMudhra certificate revocation lists. Browser handling differs: Firefox’s CRLite blocks the certificates, while Chrome, Edge, Android, and most Safari clients load them because their revocation coverage differs.
The report says Mozilla and Chrome distrust the emSign ECC Root CA - G3, with eMudhra estimating that about 30,000 certificates require migration. Its observations are time-specific and independently uncorroborated in the supplied evidence.
Hugging Face releases Trainium model-support lookup Space
Hugging Face has published the torch-neuronx model support Space, which reports whether selected Transformers Hub models run unmodified on AWS Trainium. It also exposes a public dataset containing results, failures, tested stack versions, and run scripts.
Each test runs a model on Trainium2, using tensor or expert parallelism when needed, then generates 32 tokens and checks the output. The Space records core count, wall time including compilation, peak device memory, and the relevant Neuron compiler and Transformers versions.
Coverage currently includes only selected text-generation and image-text-to-text models that fit on four Trainium2 devices. A PASS does not establish throughput or longer-generation performance.
A twelve-year history of distributed SQLite architectures
A first-person retrospective follows distributed SQLite from rqlite’s 2014 design through a broader ecosystem of projects and services. The article describes rqlite as placing a Raft log before SQLite and replicating writes across a cluster.
It distinguishes three approaches: consensus for leader-based strong consistency, WAL streaming for single-node disaster recovery, and CRDT-based merging for multiple writers with eventual consistency. It also contrasts embedded libraries and self-hosted servers with managed platforms.
The comparison is useful for selecting an architecture based on consistency, recovery, and deployment needs. Its historical and market conclusions are the author’s account, not independently verified evidence.
Moonshot releases Kimi K3.1 Pro through its API
Moonshot has released Kimi K3.1 Pro through the Kimi API, with access also reported through Kimi.ai, Kimi Work, and Kimi Code. A Highspeed variant is available alongside it, while full model weights are scheduled for release by November 4, 2026.
Moonshot describes Highspeed as two to three times faster than K3.1 Pro. The primary report lists Kimi K3.1 Pro at $2 per million input tokens and $10 per million output tokens, and notes compatibility with an OpenAI-style endpoint.
The supplied material reports strong comparative benchmark results and an Artificial Analysis score, but provides no independent methodology. Engineers should validate quality, latency, and cost for their workloads.
Wordfence reports Backuply SQL injection risk across 600,000 sites
Wordfence reports a second-order SQL injection vulnerability in the Backuply WordPress plugin, which has more than 600,000 active installations. The supplied advisory does not identify a fixed release, so the lifecycle status remains unknown.
According to Wordfence, an authenticated attacker with subscriber-level access or higher can store a malicious SQL payload. That payload is executed when an administrator later restores a database backup, creating a path to administrator privilege escalation and site takeover.
Operators should assess Backuply deployments and the database-restore workflow against this documented attack path. The supplied material does not include a CVE, exploitation evidence, or remediation version.
PHP RCE embargo breach prompts apology and disclosure debate
A Hacker News poster apologized for discussing upcoming PHP remote-code-execution vulnerabilities associated with CVE-2026-109618 and CVE-2026-109619. They said they were unaware of an embargo and intended to alert people who follow PHP’s release cycle.
The poster says the disclosure affected the coordinated-release schedule and created additional workload. Discussion participants dispute whether the poster, upstream contacts, or other parties were responsible for the embargo breach, with some arguing that information had already spread.
For security teams, the thread illustrates the operational risk of sharing vulnerability details without complete disclosure context. It provides no official technical analysis, affected-version guidance, or confirmed remediation information.
WordPress Contributor Toolkit 2.0 released with redesigned interface
WordPress Contributor Toolkit 2.0.0 has been released with a redesigned interface based on the WordPress Design System. It supports Core and Gutenberg contribution workflows and ships in ten languages.
The release adds centralized settings for themes, language, server behavior, site PHP versions, debug flags, and account details. It also presents changed files as a tree, supports per-file reverts, preserves Trac details between launches, and opens the relevant database in Adminer for randomized SQLite storage or custom DB_PATH values.
Version 2.0 is about 350 MB smaller once installed after removing unused components. The supplied material documents features and packaging changes, but not independent impact measurements.
Wordfence publishes weekly WordPress vulnerability report
Wordfence Intelligence published a weekly WordPress vulnerability report covering October 12 through October 18, 2026. The report is presented as a resource for reviewing vulnerabilities disclosed during the period and entries added to its vulnerability database.
Wordfence directs site operators to review the listed vulnerabilities to determine whether their sites are affected. The supplied excerpt does not describe individual vulnerabilities or technical remediation.
The report can serve as a dated input to WordPress exposure reviews, but the excerpt is truncated. It provides no vulnerability counts, affected components, severity information, CVE identifiers, or upgrade guidance.
WordPress 7.1 retrospective identifies release-process improvements
The WordPress 7.1 retrospective consolidates feedback from survey responses and comments about the release cycle. The post presents recommendations for future releases rather than implementation action items.
Contributors highlighted release-squad coordination, earlier communication of user-facing editor changes, documentation ownership, wider testing with hosts, and more automation. They also recommended considering the impact of tying a release to a live event and improving time-zone overlap.
The retrospective reports that half of respondents had not served on a release squad but wanted to try. It does not document adopted changes or technical implementation, so teams should treat the findings as planning input.
WordPress releases @wordpress/viewport 7.0.0
The official WordPress Gutenberg repository published @wordpress/[email protected] on October 20, 2026. The release page identifies commit 8d30f92, confirming a released package version.
The supplied page does not expose a changelog, implementation summary, migration guidance, or usage details. Its captured content consists largely of the package name, version, repository metadata, and a loading-error message.
Gutenberg and block-editor developers can verify that the package release exists, but the available evidence is insufficient to assess behavioral changes, compatibility impact, or upgrade requirements.