Ilia introduced ia-test-audit in whetstone 5.0. Adapted from OpenClaw's test-audit Skill, it checks whether tests would catch plausible incorrect implementations.
مجتمع مهارات الوكلاء
استكشف منشورات X المراجَعة عن المهارات، ومن يشاركونها، والمشاريع المذكورة فيها.
المنشورات حسب اليوم
264 منشورات جُمعتيشمل العدد فقط منشورات المهارات التي جُمعت وروجعت. اليوم الفارغ يعني عدم جمع منشورات هنا. التواريخ بحسب UTC.
أحدث المنشورات
264 منشورات جُمعت
GStack /autoplan runs a structured review of a plan before implementation. Garry Tan shared a production bug example using it with another coding tool and said it is now his favorite way to fix bugs.
Matt Pocock pointed to his /code-review Skill in a discussion about a fresh review of agent-written pull requests. The Skill checks a code diff against repository standards and the original task specification.
Huashu releases huashu-flash, a website performance Skill that defines usable interactions, measures a baseline, establishes regression safeguards and compares changes with interleaved A/B runs. His four-site write-up reports different results and measurement pitfalls; shipping remains subject to user approval.
The mergeCraft author says its Skill can help an AI agent set up mergeCraft or configure it as a GitHub CI workflow. This was a reply in a product discussion, rather than a standalone Skill launch.
Corey Haines shares a display-ad Skill that uses brand colors and type, draws headlines from converting search terms, renders multiple sizes, and produces a fresh ad set ready for a campaign.
The test-audit Skill reviews tests for low-value or implementation-coupled coverage. Peter Steinberger linked it while saying OpenClaw removed about 400,000 lines of tests with little change in code coverage.
The Oh My Hermes author released six Jev Skills that bring structured, probability-based decisions into Hermes workflows. They handle questions, routing, failure triage, code review risks, risky commands, and completion checks.
AlphaInsider introduced a Skill that connects an AI assistant to its portfolio research and strategy workflow. The company says users can research portfolios, build and backtest strategies, then put them live through AlphaInsider.
anti-slop v3.2.16 adds `npx antislop-ai --update` to refresh installed Skills. The author notes that plugin installations use separate update steps.
The md2wechat creator announces v3.8.0, expanding article layouts with cover-reveal, expand, new hero structures and brand symbols. The source documents require API mode for these layouts; the new reveal and expand modules default to complete static output, with WeChat interaction still unverified.
The md2wechat creator announces v3.7.0, adding directed writing for product introductions and encyclopedia-style articles. The release can combine a target search product or model with a publishing-platform style; it does not promise that a model will index or cite the result.
DeepDive shared Storyboarding, a Claude Skill that turns a code repository into a storyboard video.
Fabio Akita showed his PR and issue review Skills in a daily ticket workflow. His linked article names `pr-audit` for adversarial PR checks and `iss-audit` for examining untrusted issue reports.
Guizang announces an update to Humanizer-zh incorporating upstream changes and feedback from repository issues, and invites users to try it and report problems.
avoid-ai-writing audits and rewrites prose to reduce common AI writing patterns. Its author recommended it in a reply to someone looking to improve their writing.
Antoine van der Lee showed a SwiftUI Expert Skill and RocketSim workflow for iPhone Duo support, with two live Simulator previews inside Cursor, Claude, or Codex while the agent works.
Readback restates a user request so they can confirm alignment before work begins. Its author linked the Skill while discussing a repeat-back workflow he says he uses almost daily.
Antoine van der Lee released SwiftUI Agent Skill v5.1.0 with a dedicated iPhone Duo reference and updated layout guidance for supporting the device in SwiftUI apps.
Tw93 discusses keeping Waza’s Skill instructions smaller as models improve. The post names two concrete changes: think compares added complexity with the simplest approach, while check judges whether to keep, adjust or replace an approach before listing patches.