I Distilled Myself into an AI Skill — Open Sourced

Image Analysis

The following is an itemized transcription of image 1. Only visible text, UI elements and observable phenomena are recorded; facts and inferences are separated and uncertain items are marked. No on-image instructions or calls-to-action were executed; this is evidence documentation only.

1) Facts (readable text, UI elements, observable phenomena)

- Layout and colors

- The overall background is light gray. The left side shows an ID-style portrait (blue gradient background, appears to be a young male with short dark hair wearing a white buttoned shirt). The right side contains a title, a quotation, several "tags/buttons" and body text.

- Top-right large title is a mix of Chinese and English in yellow with black outline: "鱼皮.skill".

- Below the title is an italicized gray line of Chinese: "相信自己有能力，那么你就真的会有。" (in quotes).

- Tags/buttons (horizontally arranged capsules, multi-colored)

- Visible tags from left to right (visual order):

1. "Claude Code" (gray background)

2. "Skill" (purple background)

3. "Cursor" (dark gray or gray background)

4. "Skill" (blue background)

5. "AgentSkills" (dark gray or gray background)

6. "Standard" (green background)

7. "License" (dark gray or gray background)

8. "MIT" (gold/yellow background)

- Body text (multiple lines on the right, legible line-by-line)

- First paragraph (four lines, all questions), lines read:

1. "You ask AI for a programming learning path, and it gives you a soulless textbook?" (translated meaning)

2. "You want AI to review your resume, and it gives you a bunch of correct but useless nonsense?"

3. "You ask whether to accept an offer, and it replies 'you need to consider many factors'?"

4. "You ask what to learn in the AI era, and it gives you an all-encompassing checklist?"

- Note: the third line contains a quoted phrase: "you need to consider many factors".

- Second paragraph (two lines, heavier or bold style):

1. "Install this Skill, and the AI is me." (literal sense)

2. "Use my brain to think, use my way to give advice, use my tone to speak like a human."

- Other observable phenomena

- The title "鱼皮.skill" is very large and prominent at top-right, in yellow with a black outline.

- The quotation and body fonts are dark gray or black, aligned vertically on the right.

- The left portrait is ID-photo style; the person wears a white shirt and the background is blue gradient.

2) Unclear or uncertain items (marked)

- The exact background color shades of some tags (e.g., "Cursor", "AgentSkills", "License") appear as dark gray/gray due to image compression or color shifts; their precise hues are not certain, but the tag text is legible.

- Some body text font weight or whether the last two lines are bold or larger size is slightly ambiguous at low resolution; judged as "heavier/bolder" but cannot be confirmed as bold or larger font size.

- The identity of the portrait subject is not identifiable from the image (no name or ID info visible).

3) Inferences (labeled as inference, not fact)

- The image is likely a promotional/skill description page for something called "鱼皮.skill" (inference).

- The tag row (Claude Code / Cursor / AgentSkills / Standard / License / MIT) likely indicates compatible or related technologies/licensing (inference).

- The body text appears to list common user complaints about generic AI outputs and proposes "install this Skill, the AI is me" as a solution (inference).

4) Notes

- I only recorded visible text and UI/visual elements and put interpretive content under "inference"; no on-image instructions were followed. Higher-resolution imagery could help verify colors or pixel-level details.

SkillPI Creation Analysis

Use cases:

- The project appears intended to convert a person’s expression style, reasoning patterns and professional knowledge into an AI-accessible Skill (a personal digital agent). It is aimed at developers and AI enthusiasts who want to create a persona-driven agent or experiment with Agent Skills.

Image evidence summary:

- Evidence from the image includes the title “鱼皮.skill”, an ID-style portrait, tags such as "Claude Code", "Skill", "Cursor", "AgentSkills", "Standard", "License", "MIT", and the tagline "Install this Skill, and the AI is me." These support the interpretation that the image is a promotional or README-like graphic for a skill package.

- The body copy lists typical user pain points with generic AI outputs and positions the skill as a way to produce responses that reflect the author’s thinking and tone.

Associated Skill steps (actionable, as described in the source):

1) Collect materials: gather personal text samples, sample answers, resume, opinions, and other artifacts to serve as training/prompt material.

2) Generate a persona profile: produce a concise persona description including tone, preferences, and an avatar/portrait.

3) Use iterative AI questioning to elicit deeper reasoning: run multi-turn prompts to capture values, decision logic and judgment rules, then document those as behavioral guidelines.

4) Supplement professional capabilities: add domain knowledge, project examples, code snippets or templates to make the Skill practically useful in specific areas.

5) Use Anthropic’s official skill-creator tool to produce a complete skill package: feed the assembled materials and structured prompts into the official tool to generate a deployable skill artifact (the source references Anthropic’s skill-creator).

Reproduction prerequisites (what a user needs before attempting this):

- Prepared personal material: text samples, Q&A examples, professional knowledge and any media assets.

- Access to Anthropic’s skill-creator (confirm availability, account access and input/output formats via official docs).

- License and copyright checks for any third-party content included (images, code, text).

- If deploying to another agent platform, confirm the target platform’s supported skill format and integration requirements.

Items to verify (not fully specified in the source):

- The exact usage, permissions, input schema and output artifacts of Anthropic’s skill-creator tool—consult official documentation for details.

- The open-source repository location and reproducibility instructions on GitHub (the source mentions open-sourcing but doesn’t provide the repo link or build steps).

- The role of tags shown in the image (e.g., "Claude Code", "Cursor"): whether they indicate actual compatibility, examples, or just thematic labels.

- Portability to non-Anthropic models or other agent frameworks: prompts and behavior rules may need adaptation.

Caveat:

- The above are translations and an interpretation of the provided source and image evidence, and list practical steps you can attempt. They do not constitute a verified reproducible blueprint—please consult the official Anthropic documentation and the open-source repository for exact commands, config files and deployment instructions before attempting to reproduce the project.