Cangjie-skill Real-world Test - Convert Books, Videos, and Podcasts into Usable Skills with One Click
Today I'm sharing some ideas on how to turn methodologies into skills. Often, after reading a book or watching a video, I feel like I've learned something, but I can't fully apply it to my own projects, wasting a lot of inspiration.
Today I'll share something suitable for turning methodologies into skills.
When I finish reading a book or watching a video, I always feel like I've learned something, but I can't fully apply it to my own projects, which wastes a lot of inspiration.
Today I'm going to give you...recommendA GitHub project called Cangjie-skill can distill the books, videos, and blogs you've read into Skills, which you can then use in real-world scenarios.
As you can see, this is the result of distilling the lighting effects tutorial video and applying it to the raw image project:
Case 1: Book distillation skill
We directly call this skill to perform distillation:
Prompt words:
Please use cangjie-skill to perform a single skill distillation on the book I provided.
Book Title: The Copywriter's Handbook
Author: Robert W. Bly
The file path is: "D:\360MoveData\Users\win\Desktop\文案创意完全手册-Robert Bly.pdf"
After distillation is complete, we'll do the same project using this skill and without using this skill:
Use the distilled skill:
Prompt words:
Please call $feature-to-benefit to complete the following task.
Please write the following product information as the landing page homepage copy.
Product: Ticket Folder, an invoice organization software designed for the finance teams of small businesses.
Target users:
A finance team of 2 to 5 people processes 300 to 1000 invoices per month.
Product Information:
1. Up to 30 electronic invoices (PDF) can be uploaded at a time.
2. Internally, 500 standard electronic invoices were used for testing, and the overall recognition accuracy rate for fields such as amount, date, and invoice code was 96.8%. Photos of paper invoices were not tested.
3. The system alerts users to duplicate invoices based on the invoice code and invoice number.
4. It can export Kingdee standard templates, but currently does not support UFIDA templates.
5. Administrators can set member permissions and view operation logs.
6. Of the 8 finance staff interviewed, 6 said that manually entering and checking duplicate invoices took up a lot of time in the invoice sorting process.
Please output:
1. A title of no more than 18 characters.
2. A subtitle of no more than 50 words.
3. Three selling point cards
4. Copy for a "Try" button
Prompt words:
Please write the following product information as the landing page homepage copy.
Product: Ticket Folder, an invoice organization software designed for the finance teams of small businesses.
Target users:
A finance team of 2 to 5 people processes 300 to 1000 invoices per month.
Product Information:
1. Up to 30 electronic invoices (PDF) can be uploaded at a time.
2. Internally, 500 standard electronic invoices were used for testing, and the overall recognition accuracy rate for fields such as amount, date, and invoice code was 96.8%. Photos of paper invoices were not tested.
3. The system alerts users to duplicate invoices based on the invoice code and invoice number.
4. It can export Kingdee standard templates, but currently does not support UFIDA templates.
5. Administrators can set member permissions and view operation logs.
6. Of the 8 finance staff interviewed, 6 said that manually entering and checking duplicate invoices took up a lot of time in the invoice sorting process.
Please output:
1. A title of no more than 18 characters.
2. A subtitle of no more than 50 words.
3. Three selling point cards
4. Copy for a "Try" button
The version without Skills has a target user base and a sense of business volume, but it has several issues: The thousands of invoices come from the target user's monthly processing volume of 300-1000 invoices, which doesn't mean the invoice wallet has been validated for handling thousands of invoices; too many functions are crammed into one place for deduplication, collaboration, and export; the buttons are poorly written...freeThe trial ticket holder is missing the original data.freepolicy.
After calling feature-to-benefit, the title becomes: Batch recognition of invoices, reducing manual data entry, linking the product feature with the benefits received by finance personnel.
The three cards handle data entry, deduplication, and Kingdee export respectively, ensuring that every benefit is traceable back to the product's actual functionality. The deduplication check is described as: "Prompt suspected duplicates," subtly implying that the system can completely eliminate duplicate reimbursements.
Case 2 Video distillation skill
We directly call this skill to perform distillation:
Prompt words:
https://www.bilibili.com/video/BV1xQnPzdEiy/ Could you please help me distill this lighting effect?
After distillation is complete, we'll work on the same project using this skill and without it:
Use the distilled skill:
Prompt words:
First, use `$prompt-lighting-director` to generate a 16:9 promotional image for a psychological horror game. Subject: A young Asian woman with short black hair, wearing a plain light gray shirt, standing in front of a gray concrete wall. The figure faces the camera directly, focusing on her chest and above, with a calm expression and no exaggerated movements.
Without skill:
Prompt words:
Generate a 16:9 promotional image for a psychological horror game.
main body:
A young Asian woman with short black hair, wearing a plain light gray shirt, stands in front of a gray concrete wall. She faces the camera directly, the shot focusing on her chest and above, her expression calm and without exaggerated movements.
The lighting effects best highlight the difference. The image without Skill uses a common top-heavy, frontal cool light, with the forehead, bridge of the nose, and cheekbones being the brightest areas, while the shadows on the nose and chin point downwards. The character's face is clear, and the gray concrete wall has texture, but overall it looks more like a portrait of someone in an interrogation room, lacking a strong sense of horror.
After calling prompt-lighting-director, the light source is placed below the character's chest. The chin, nostrils, and lower lip become the brightest areas, while the nose shadow, cheekbone shadow, and eye shadow abnormally move upwards, creating a sense of oppression entirely from the direction of the shadows.
Case 3: Podcast Distillation Skill
We directly call this skill to perform distillation:
Prompt words:
Please call cangjie-skill to distill the following podcast into a Skill.
https://podscripts.co/podcasts/acquired/superhuman-with-ceo-rahul-vohra
After distillation is complete, we'll do the same project using this skill and without using this skill:
Use the distilled skill:
Prompt words:
Please call $superhuman-pmf-engine to complete the following task.
We are testing a AI Meeting minutes product. Please determine if the product has been obtained. PMFWho should be prioritized for service, and what should be developed in the next six weeks?
Sample rules:
All respondents had completed at least two real meetings within the past 14 days and had used the action item extraction feature. There were no duplicate responses or missing data.
A total of 48 valid questionnaires were submitted.
Founders, 12 people:
– Very disappointed (8 people)
– Somewhat disappointed (3 people)
– Not disappointed (1 person)
Sales Managers, 12 people:
– Very disappointed (7 people)
– Somewhat disappointed (4 people)
– Not disappointed (1 person)
Product Manager, 12 people:
– Very disappointed (2 people)
– Somewhat disappointed (5 people)
– Not disappointed (5 people)
Independent consultants, 12 people:
– Very disappointed (1 person)
– Somewhat disappointed (5 people)
– Not disappointed (6 people)
The main benefits mentioned by the 18 "very disappointed" users:
– Action items accurate: 13 people
– Meeting minutes are searchable for: 3 people
– Others: 2 people
Of the 17 users who expressed "some disappointment," 8 also acknowledged that "the action items were accurate."
– 5 people were primarily hindered by calendar synchronization issues.
– Two people were primarily hampered by CRM synchronization issues.
– 1 person needs to customize the export format
Concurrent behavioral data:
– Founder's four-week retention rate: 75%; paid conversion rate: 50%.
– Sales managers achieved a 67% four-week retention rate and a 42% paid conversion rate.
– Product manager's four-week retention rate was 33%, and the paid conversion rate was 17%.
– Independent consultants have a 25% four-week retention rate and an 8% paid conversion rate.
– Only 24% of all registered users completed both core actions for activation.
Please output:
1. PMF Judgment
2. Prioritize target users
3. Roadmap for the next six weeks
4. Should we expand customer acquisition?
5. Next round of investigation plan
Prompt wordsWe are testing a AI Meeting Minutes Product. Please determine whether the product has already obtained PMF (Progressive Meeting Decision), who should be prioritized for it, and what should be developed in the next six weeks. Sample Rules: All respondents had completed at least two real meetings within the last 14 days and had experienced the action item extraction function. There were no duplicate responses or missing data. 48 valid responses were received: Founders, 12 individuals.
– Very disappointed (8 people)
– Somewhat disappointed (3 people)
– Not disappointed (1 person)
Sales Managers, 12 people:
– Very disappointed (7 people)
– Somewhat disappointed (4 people)
– Not disappointed (1 person)
Product Manager, 12 people:
– Very disappointed (2 people)
– Somewhat disappointed (5 people)
– Not disappointed (5 people)
Independent consultants, 12 people:
– Very disappointed (1 person)
– Somewhat disappointed (5 people)
– Not disappointed (6 people)
The main benefits mentioned by the 18 "very disappointed" users:
– Action items accurate: 13 people
– Meeting minutes are searchable for: 3 people
– Others: 2 people
Of the 17 users who expressed "some disappointment," 8 also acknowledged that "the action items were accurate."
– 5 people were primarily hindered by calendar synchronization issues.
– Two people were primarily hampered by CRM synchronization issues.
– 1 person needs to customize the export format
Concurrent behavioral data:
– Founder's four-week retention rate: 75%; paid conversion rate: 50%.
– Sales managers achieved a 67% four-week retention rate and a 42% paid conversion rate.
– Product manager's four-week retention rate was 33%, and the paid conversion rate was 17%.
– Independent consultants have a 25% four-week retention rate and an 8% paid conversion rate.
– Only 24% of all registered users completed both core actions for activation.
Please output:
PMF Judgment
Priority target users
Roadmap for the next six weeks
Should we expand customer acquisition?
Next round of investigation plan
The version without Skills directly labeled the results of each group of 12 people as the strongest PMF and strong PMF, without noting that the sample size of the clusters was small. Information that was not provided was also included in the founder profile.
The superhuman-pmf-engine retains the denominator for each decision. Skill also confirmed the shared interest in action item accuracy from 18 strong supporters, and then filtered out 8 users from 17 somewhat disappointed users who also recognized this interest. The roadmap outlines the action item accuracy, calendar synchronization, and activation process.
Cangjie-skill It is a set of "long content conversion" AI The methodology of "Skill".Cangjie-skill Read books, video subtitles, podcast transcripts, courses, or interviews, break down reusable methodologies into multiple independent skills, and assign them to... AI Agent Call it in specific problems.
Project address: https://github.com/kangarooking/cangjie-skill
Cangjie-skill will solve severalpracticalQuestion:
- Which viewpoints qualify as independent skills?
- What does the user say?Agent The corresponding Skill should be invoked;
- Agent What steps should be executed after the call?
- In what situations should we stop or switch to other methods?
- How to combine multiple skills and how to avoid them competing for calls?
Cangjie-skill More suitable for:
- Business, investment, management, and technology courses with a high methodological density;
- Internal team training materials and expert interviews;
- Hope to establish a personal Agent Tool library researchers;
- Content distillation requires the preservation of source, obsolescence records, and method boundaries.
To use this project, just let AI Agent I'll install it for you. Once it's installed, you can start enjoying your favorite books, videos, or podcasts.
In the past, after reading a good book or watching a video, what remained was often a collection, notes, and a vague impression. When I actually encountered a project, I had to go through the materials again, and many useful methods were gradually forgotten.
Cangjie-skill provides a better way to present good content.practicalThe methods for saving these elements: copywriting techniques from books, lighting tips from videos, and product judgment frameworks from podcasts can all be broken down into Skills, along with their applicable conditions, execution steps, and usage boundaries.
The next time you write product copy, design lighting effects, or judge PMF (Product Motion Factor),Agent You can directly call the corresponding Skill and work using the same set of methods. Individuals can avoid the detours of repetitive research, and teams can continue to use the experience left by experts.
When you come across books, videos, or podcasts that are worth studying repeatedly, you can take an extra step: turn the most useful methods into skills.
Bookmarks preserve content, Skills preserve the experience within that content, allowing good content to continue to be useful in one real project after another.
Original link:Achieve a whopping 4.2K stars on GitHub, turning all knowledge into skills.