10 Essential AI Skills I Use Daily (Don’t Hoard Unneeded Skills)

1. Web Capture – Dokobot

Dokobot is a free browser‑plugin and CLI skill that reliably extracts web page content. According to its documentation it can reduce token consumption by up to 99 % compared with raw HTML. It supports scrolling long pages, handles lazy‑loaded elements, and performs visual extraction, which often simplifies processing of complex layouts. The author notes that as of August 2026 no usage‑limit has been encountered, though official limits may apply.

2. Deep Article Reading – article‑interpreter

This skill reads high‑quality articles, papers, or reports and produces a multi‑layered output: a concise summary, key points, personal insights, critical analysis, and extended discussion. It tailors the analysis to the user’s research direction or job role, and highlights possible errors or weak arguments. Installation is done by sending a prompt containing the GitHub URL to the Agent.

3. Local Audio/Video Transcription – local‑audio‑transcriber

When batch transcription limits or fees become a bottleneck, this skill runs entirely on the local machine, converting MP3 audio or MP4 video into text. The author provides a ready‑to‑use prompt for installation, after which the user supplies the media file to the Agent and specifies the output directory.

4. Skill Review & Optimization – skill‑optimizer

Over time a collection of Skills can develop rule conflicts, structural chaos, or unclear trigger conditions. skill‑optimizer scans a Skill, diagnoses issues, matches appropriate design patterns (inspired by Anthropic best practices and Google’s Skill design patterns), and proposes a prioritized modification plan. It reports problems without automatically applying changes, allowing the user to confirm the plan before execution.

5. Engineering Method Library – mattpocock/skills

This GitHub repository, with over 210 k stars as of August 2026, bundles reusable engineering methods for Agents. Skills are classified as User‑invoked (explicitly triggered for workflow or decision making) or Model‑invoked (automatically called for TDD, diagnostics, or domain modeling). Installation uses npx skills@latest add mattpocock/skills followed by project‑specific setup via /setup-matt-pocock-skills. The author advises against installing the whole set for a single problem.

6. WeChat Article Formatting – gzh‑design‑skill

gzh‑design‑skill converts Markdown into HTML that fits the WeChat editor, offering six built‑in themes. It streamlines the layout of articles containing code blocks, quotes, images, and multi‑level headings, saving considerable manual formatting time. Users can further customize themes to match personal aesthetic preferences.

7. Product Video Generation – Video‑ShotCraft

Video‑ShotCraft is an AI Agent skill that turns product assets into dynamic promotional, release, or demo videos using Remotion. It handles storyboard design, animation, sound design, and can embed real‑page recordings, 2.5D camera moves, and cinematic sound effects. The repository has amassed over 4.3 k stars on GitHub.

8. Automatic Editing of Spoken Videos – ChatCut

ChatCut focuses on automatic cutting of voice‑over videos. It works alongside Video‑ShotCraft and can be combined with a subtitle‑correction skill to address subtitle errors for technical terms or names. The author notes that while the results are close to expectations, some fine‑tuning may still be required.

9. Batch Photo Download – alltuu‑downloader / photoplus‑downloader

These two skills automate bulk downloading from common event‑photo platforms (alltuu/m.alltuu and PhotoPlus). Instead of manually clicking each image, the user sends the album link to the Agent and specifies a target directory. Installation is performed via prompts that reference the respective GitHub URLs.

10. Skill Discovery – find‑skills

When a specific Skill is needed, find‑skills searches the Skill marketplace based on a natural‑language problem description, presents suitable candidates, and installs the chosen one after user confirmation. The author demonstrates installation with a prompt that runs

npx skills add https://github.com/vercel-labs/skills --skill find-skills

.

Overall, the article emphasizes selecting Skills that directly address real problems, avoiding unnecessary installations, and leveraging the described evaluation criteria—such as token efficiency, star count, local execution, and compatibility with existing workflows—to build an effective AI‑augmented toolkit.