vincelele/ai-fomo-skills/ai-fomo-init

AI FOMO Init

Purpose

Create a local Personal Alignment Layer and starter workspace. This skill aligns the agent before any source is summarized, collected, or filed.

Use When

Use this skill when the user asks to:

• set up AI FOMO
• initialize a personal AI alignment workspace
• create an anti-FOMO knowledge workspace
• configure what AI information is worth attention
• generate starter folders and templates

Do Not Use When

Do not use this skill to:

• summarize a specific source
• collect from a source
• write wiki/sources
• generate a digest from existing material

Use ai-fomo-sources for source intake and ai-fomo for judgment and filing.

Core Workflow

1. Choose or confirm the target workspace path.
2. Ask the user for a profile brief: resume, bio, portfolio, LinkedIn-style summary, project history, or free-form notes.
3. Extract a first-pass alignment map from the profile brief.
4. Ask targeted QA to fill only the gaps that matter for AI information filtering.
5. Copy the starter workspace from assets/starter-workspace/ if the user wants files created.
6. Write confirmed alignment context into self-context/.
7. Mark weak inferences as draft or pending confirmation.
8. End with a short alignment contract summary for the user to review.

Two-Stage Alignment Intake

Stage 1: Profile Brief

Invite the user to provide one or more of:

• resume or CV
• personal bio
• project list
• portfolio notes
• work history
• current goals
• examples of AI content they found valuable or useless

From this, extract only tentative alignment signals:

• role and current work
• long-term themes
• recurring decisions
• technical or product interests
• likely source preferences
• possible downrank patterns

Do not treat these as final until confirmed.

Stage 2: Targeted QA

Ask a small number of focused questions to close gaps. Prefer 3 to 7 questions.

Question areas:

• what decisions the user wants AI information to improve
• which AI topics are high priority now
• which topics are interesting but lower priority
• what kinds of content waste the user's time
• what sources they trust or distrust
• how aggressive the agent should be about skipping material
• what output style is most useful

Do not ask a long onboarding survey. The goal is enough alignment to start, then improve through feedback.

Personal Alignment Layer

The minimum layer should answer:

• who the agent is serving
• what the user is building, studying, or deciding
• which AI themes are high priority
• which content types should be downranked
• what counts as high-signal information
• how feedback should change future filtering

If the user provides too little information, ask concise follow-up questions instead of overfitting.

Output Contract

When initialization finishes, report:

• workspace path
• files created or updated
• confirmed alignment facts from the profile brief and QA
• draft assumptions still requiring user review
• next recommended action: connect sources or process the first source

Safety Rules

• Do not write credentials into the workspace.
• Do not copy private examples into templates.
• Do not infer sensitive personal context unless explicitly provided.
• Do not assume the user's interests match the author's interests.
• Do not treat one-time task goals as long-term preferences.
• Do not turn resume details into public examples or shareable templates.