中 Download Now

Advanced Recipes

SailFish’s skills load automatically based on your needs, and multiple skills can chain together seamlessly. Just describe your end goal, and the AI will orchestrate all necessary capabilities to get it done.

Here are some real-world examples of multi-skill workflows to spark your imagination.

Resume Screening & Sorting

Skills involved: PDF / file reading + Excel + Word

Scenario: You’ve received a batch of resumes (PDFs or scans) and need to quickly screen for qualified candidates and compile a report.

I have 20 resumes (PDF) in ~/Desktop/Resumes/.
Screen for candidates with 3+ years of Java backend experience.
Compile everyone's key info (name, years of experience, tech stack, most recent company) into an Excel sheet.
Then pick the top 5 matches and generate a Word recommendation report with reasons.

The AI will:

  1. Read each PDF resume and extract key information
  2. Filter by criteria and generate an Excel summary
  3. Select the best candidates and produce a Word report with recommendations

You can follow up:

Send the recommendation report to the HR lead via Feishu.

Competitive Analysis Report

Skills involved: Browser + Excel + Word

Scenario: Research competitor products and compile a structured analysis.

Help me create a competitive analysis:
1. Search for the top AI terminal/agent products (at least 5)
2. Collect each product's core features, pricing, and target users
3. Organize into a comparison Excel spreadsheet
4. Write a competitive analysis Word report with feature comparison, strengths/weaknesses, and differentiation suggestions

Automated Data Reporting

Skills involved: Terminal + Excel + Word + Email + Awaken Mode

Scenario: Periodically collect data from servers, generate reports, and distribute them.

Set up an automated task:
- Run every Monday at 9 AM
- Log into the production server and collect last week's API call volume, error rates, and response times
- Organize the data into an Excel sheet with trend comparisons
- Generate a "Weekly API Quality Report" Word document
- Email it to tech-team@company.com

The AI will create an Awaken Mode watch that automatically executes the entire workflow on schedule.

Document Translation & Localization

Skills involved: PDF / Word + Word

Scenario: Translate technical documents or contracts while preserving formatting.

Read ~/Documents/API-Specification.pdf,
translate the content to Chinese while preserving the original chapter structure,
and generate a Chinese version as a Word document.
Keep technical terms in English with Chinese explanations in parentheses.

Or batch translation:

Translate all English Markdown files in ~/docs/ to Chinese.
Save translated files with the same names under ~/docs/zh/.

Email-Driven Workflows

Skills involved: Email + Calendar + Excel + Word

Scenario: Automatically process specific emails, extract information, and take follow-up actions.

Please do the following:
1. Check my inbox for emails from *@supplier.com in the last 3 days
2. Extract pricing info from each (product name, unit price, quantity, delivery date)
3. Compile into a price comparison Excel sheet
4. If any quotes are under our budget (reference ~/budget.xlsx), create a follow-up reminder on my calendar
5. Generate a procurement recommendation Word document

Web Data Collection & Analysis

Skills involved: Browser + Excel + Word + Email/IM

Scenario: Scrape data from websites for analysis.

Collect data from Hacker News:
1. Get today's top 30 stories
2. Record title, author, points, number of comments
3. Organize into Excel sorted by points
4. Write a brief trend analysis and send it to Slack

Automated Ops Daily Report

Skills involved: Terminal + Knowledge Base + Word + IM + Awaken Mode

Scenario: Automatically collect status from multiple servers, generate daily reports, and push notifications.

Create a watch:
- Run daily at 5:30 PM
- Check these 3 servers (SSH already configured):
  web-server, db-server, app-server
- Collect: CPU/memory usage, disk space, service status, today's error logs
- Compare with yesterday's data (from knowledge base)
- Generate an Ops Daily Report Word document
- If anomalies found, notify me immediately via DingTalk; if all normal, send the report to the WeCom ops group
- Save today's data to the knowledge base for future comparison

End-to-End Meeting Automation

Skills involved: Calendar + Email + Word + IM

Scenario: Automate the full meeting lifecycle from scheduling to minutes.

Before the meeting:

Check my schedule for tomorrow, find all meetings,
and send a reminder email to each meeting's attendees with the agenda.

After the meeting:

Help me create minutes for this afternoon's product review:
- Decided to delay v2.0 launch to end of March
- Adding "Data Export" feature, assigned to Wang Ming
- Design mockups due by next Wednesday
Create a Word minutes document, send it to all attendees via Feishu,
and create calendar reminders for the key milestones.

Technical Research & Knowledge Building

Skills involved: Browser + Knowledge Base + Word

Scenario: Research a technical topic, compile documentation, and save to the knowledge base.

Research Kubernetes HPA (Horizontal Pod Autoscaler):
1. Search and read official docs and best practices
2. Understand configuration methods, metric types, and common pitfalls
3. Write a technical research document (Word) with: principles, config examples, production tips
4. Save key takeaways to the knowledge base for quick reference later

Tips for Combining Skills

Use Natural Language

No special syntax required—just describe the complete workflow. The AI will figure out which skills to load and in what order.

Step-by-Step Works Too

If describing everything at once feels overwhelming, give instructions one at a time:

First, read all CSV files in ~/data/

Summarize the data by month and create an Excel file

Write an analysis report based on the Excel data and email it

Each step’s results stay in the Agent’s context, so subsequent steps can naturally reference earlier work.

Leverage Awaken Mode

If a workflow needs to run on a schedule, set it up as a watch. The AI will automatically execute the entire pipeline at the specified time—you just review the results.

Leverage the Knowledge Base

The knowledge base carries information across conversations. Save screening criteria, document templates, or common configurations there, and the AI will automatically reference them during tasks.

Works Remotely Too

All these workflows can be triggered remotely via DingTalk, Feishu, WeCom, Slack, or Telegram. Fire off a message from anywhere and let the AI handle the rest.