DEV Community: 郭帅

From chat output to deliverable files: three agent skills for office work

郭帅 — Thu, 05 Nov 2026 06:00:51 +0000

An answer in chat can give you the right content and still leave you without a file a teammate can open, check, and reuse. Start with the format you need to hand over. Word, spreadsheets, and slides call for different inputs and different checks.

Word documents with docx

Anthropic’s docx skill covers reading, creating, and editing Word files. Provide the audience, source notes, heading structure, and a reference format. For a project report, ask for a short summary, a decision table, and a section for missing information. Open the finished document to check heading levels, page breaks, and tables. Verify that each factual claim came from the supplied material.

Recalculable spreadsheets with xlsx

Use the xlsx skill when a spreadsheet is the primary deliverable. Provide the workbook, field definitions, and rules for grouping or excluding rows. For a monthly sales summary, preserve the source data and put the calculations in a separate sheet. Recalculate the workbook, inspect the formula ranges, and check a few totals against the original rows.

Slides that fit the template with pptx

The pptx skill is for work involving PowerPoint decks. Start with the audience, the decision you need from them, and the template. Review a slide outline first. Once the deck is generated, open it and inspect each slide for text overflow, crowded charts, and missing or unreadable labels.

These are example briefs and acceptance checks. The three upstream skills have their own license terms. Check the terms and required tools for your host before trying one, then use a small sample you can inspect yourself.

Adapted from the Skillπ team’s Chinese article on Juejin. This English adaptation was generated with AI.

Three agent skills for researching, editing, and illustrating a post

郭帅 — Thu, 05 Nov 2026 06:00:33 +0000

Research, editing, and illustration produce different kinds of evidence. Keeping them separate makes it easier to spot a weak citation, a changed claim, or a media file that will fail on the destination platform.

Start with a source table

Composio’s content-research-writer organizes research, outlining, citations, and feedback. Give it the topic, audience, and sources you trust. Ask for a small table with each proposed claim, its source, and anything that still needs checking. Open the primary sources before using a statistic or quotation.

Edit an existing draft against the source table

Humanizer offers an editing workflow for generic phrasing and repetitive structure. Supply a paragraph you wrote and a short sample of your voice. Ask it to remove vague wording while preserving the facts and citations. Compare the before and after for changed numbers, stronger claims, or invented examples. Treat the output as an edit you still need to review.

Add a small animation when it serves the post

Anthropic’s slack-gif-creator focuses on making GIFs for Slack. For a simple celebration animation, specify the loop, dimensions, and destination. Open the exported file, watch a few loops, and verify the file against the destination’s requirements. Check the other platform’s limits separately if you reuse it elsewhere.

For a first run, keep the source table, the edited draft, and the exported GIF as separate artifacts. Read each skill’s upstream instructions and license and check the tools your agent needs before running it.

Adapted from the Skillπ team’s Chinese article on Juejin. This English adaptation was generated with AI.

Three agent skills for a brief, a poster, and a consistent visual style

郭帅 — Thu, 05 Nov 2026 05:54:55 +0000

A small event makes a useful first exercise for agent skills. You need a one-page brief, a poster, and a style you can reuse in follow-up materials. Give each step its own input and check its output before moving on.

Turn rough notes into a brief

Use Anthropic’s doc-coauthoring to structure a document with your agent. Supply the audience, goal, raw notes, and constraints. For a book club, ask for a one-page proposal with a clear purpose, a rough agenda, and open questions. Confirm the date, venue, and other details from your notes before treating the brief as final.

Make a static poster from the approved brief

Anthropic’s canvas-design describes a workflow for static artwork in PNG or PDF. Give it the exact title and event details, the intended dimensions, and a visual direction. Open the exported file and check text, font rendering, spacing, and readability at the intended display size.

Carry the same style into a second asset

Anthropic’s theme-factory provides preset visual themes. Inspect the examples and choose a palette and typography before styling a second artifact, such as a one-page event recap. Compare both outputs side by side for consistent headings, contrast, and margins.

Before running any skill, read its current instructions and license and check that your host supports the required tools. The book club example is an exercise you can try. It describes the checks to perform on your own outputs.

Adapted from the Skillπ team’s Chinese article on Juejin. This English adaptation was generated with AI.

How to choose an AI agent skill for a real task

郭帅 — Thu, 05 Nov 2026 05:52:33 +0000

A useful agent skill starts with a job you can describe and a result you can check. Consider a local web app where a sign-in button sometimes stops working. The job is to reproduce the failure and capture evidence a teammate can use. That gives you a better search query than “web development” and a clear way to evaluate the result.

What an agent skill does

An agent skill packages reusable instructions for a task. It may also include scripts, templates, or reference material. For browser testing, that might mean instructions for starting a local app, inspecting the rendered page, and collecting a screenshot or browser logs. Your agent and its tools still do the work, so the host environment and available permissions matter.

Skillπ is the AI agent skills marketplace maintained by the team behind this DEV organization. Its catalog lets you look up skills by task or browse categories such as Testing & security. Listings can point you to the original author, source, version, license, and setup requirements. Those details help you decide what to inspect upstream.

A worked example with a local browser test

Describe the outcome. Search for “test a local web app” or “Playwright browser testing.” For the sign-in example, write down what counts as a pass before choosing a skill. An invalid test login should show an error and keep the user on the sign-in page.
Check the fit. Anthropic’s webapp-testing skill is one candidate. Its source describes a Python and Playwright workflow. Check that your agent can run those tools, that you can start the app locally, and that the license covers your intended use. If you need CI reports or coverage across several browsers, check those requirements separately.
Read the upstream instructions and scripts. Check the pinned revision and what the helper scripts launch, read, or write. A source review, a scan, and a successful sample run each tell you something different. Match the available evidence to the environment you plan to use.
Try one bounded run. Use a local copy with synthetic credentials. Ask the agent to reproduce the one flow, record the observed behavior, and return a screenshot or the relevant browser log. Review that evidence before widening the task or granting more access.

Here is a sample brief you can adapt after reviewing the skill:

In the local test app, submit an invalid login using a synthetic account. Verify that an error is shown and that the page stays on the sign-in screen. Report the steps, the expected and observed behavior, and the location of the screenshot. Use only the local test environment.

The same approach works for a spreadsheet, a code review, or a research task. Define the output first, check the skill’s requirements against your environment, and use a small run to see whether it delivers what you need.

Disclosure: This article was written with AI and published by the Skillπ team’s DEV organization.