Proposal & Cover Letter Discovery Interview

This is NOT a CV builder. This skill creates high-quality, reusable statements for proposals and cover letters by running a structured interview to extract personal-but-professional facts.

The Problem with Generic Letters

Generic cover letter and proposal writers are bland because they don't know YOU. They can't capture:

• Your living in Dubai and Seoul, and what that taught you about cross-cultural collaboration
• The robotics project you tinkered with at home that shows your passion
• The class that changed how you think about systems
• The boring job that taught you what quality really means
• Your unfair advantage that clients actually hire you for

This skill solves that through interview → facts → statements → approval → assembly.

How It Works

Phase 1: Discovery Interview

The skill asks structured questions to extract:

• Geographic and cultural context (places lived, travel, cross-cultural work)
• Work experience beyond the resume (lessons, standout moments, what colleagues rely on you for)
• Education (specific classes, formative learning, projects)
• Projects and proof (side projects, demos, artifacts)
• Awards and recognition (scholarships, competitions, certifications)
• Past hobbies (skills hidden in former interests)
• Books and influences (working philosophy)
• The boring stuff (what you learned from frustrating work)
• Proposal leverage (what you want to be hired for)
• Constraints (topics to avoid, boundaries)

Phase 2: Statement Generation

From gathered facts, the skill drafts 3-8 candidate statements per round. Each statement:

• Is grounded in real facts (no invention)
• Is usable in proposals/cover letters
• Has multiple variants (Upwork short, standard cover letter, technical/proof-first)
• Has tags (#leadership #robotics #global-living)

Phase 3: Approval Loop

You review each statement: Approve / Edit / Reject

• Approved statements go into statements.md
• Edits refine the statement and update preferences
• Rejections inform future drafting

Phase 4: Assembly (Future)

When you need a proposal, the skill pulls relevant approved statements and assembles them for the target platform and company.

Storage Model: Folders + Files

All data is stored INSIDE the skill folder at: skills/proposal-interview/personal/ and skills/proposal-interview/companies/

Directory Structure

skills/proposal-interview/
  personal/
    <person_id_or_name>/
      profile.md           # Current snapshot: name, what they do, where, objectives (ALWAYS READ FIRST)
      user.md              # Raw facts about this person (APPEND ONLY - NEVER DELETE)
      statements.md        # Approved statements (APPEND ONLY - NEVER DELETE APPROVED STATEMENTS)
      preferences.md       # Writing style preferences
      coherence.md         # Conflicts, gaps, clarifications needed

  companies/
    <company_or_initiative_slug>/
      profile.md           # Current snapshot: company name, what they do, where, objectives
      org.md               # Company facts, domain, relationships (APPEND ONLY - NEVER DELETE)
      statements.md        # Company-specific approved statements (APPEND ONLY - NEVER DELETE)
      preferences.md       # Company-specific style preferences
      projects.md          # (Optional) Project history with this org
      coherence.md         # (Optional) Company-specific conflicts/gaps

Routing Rules

Personal vs Company:

• Personal facts → skills/proposal-interview/personal/<person>/user.md
• Company-specific facts → skills/proposal-interview/companies/<slug>/org.md

Multi-person support:

• Default: current user
• If doing this for someone else (e.g., spouse), use skills/proposal-interview/personal/<their_name>/

Relationships:

• If a company has an owner/founder relevant to letters, record in org.md

File Rules (Append-Only)

⚠️ CRITICAL FOR ALL MODELS - READ THIS:

NEVER DELETE OR REWRITE LINES IN ANY FILE. ALWAYS APPEND. NEVER DELETE. ONLY ADD.

This applies to:

• ❌ NEVER delete approved statements from statements.md
• ❌ NEVER delete facts from user.md or org.md
• ❌ NEVER rewrite existing lines
• ✅ ALWAYS append new information
• ✅ ALWAYS read existing files before appending

If a conflict is detected:

1. Append the new info (don't delete the old info)
2. Append a note to coherence.md
3. Ask a clarifying question next round

profile.md is the EXCEPTION - this file CAN be edited/rewritten as it's a current snapshot, not a permanent record.

profile.md (personal or company) — ALWAYS READ FIRST

Current snapshot of the person or company. Simple, clear, updated format.

Personal profile.md format:

# [Full Name]

## What They Do
[Simple description - "AI consultant specializing in OpenClaw implementations and AI strategy"]

## Where Based
[Location - "Amsterdam, Netherlands" or "Remote" or "San Francisco, CA"]

## Primary Objective
[Main goal for proposals/applications - "Secure OpenClaw implementation contracts with mid-size tech companies"]

## Secondary Objective (Optional)
[Secondary goal - "Build reputation as AI strategy consultant for startups"]

## Current Status
[Brief context - "Actively applying to Upwork projects and direct outreach to companies"]

## Last Updated
[Date - "2026-02-24"]

Company profile.md format:

# [Company Name]

## What They Do
[Simple description - "Industrial automation, warehouse robotics"]

## Where Based
[Location - "San Francisco, CA"]

## Primary Relationship Objective
[Why we're tracking them - "Potential client for OpenClaw implementation"]

## Secondary Objective (Optional)
[Additional context - "Investor prospect for Way Back Project"]

## Key Contact
[Name and role if known - "Jane Doe, CTO"]

## Last Updated
[Date - "2026-02-24"]

Workflow: Always read profile.md FIRST (before user.md, statements.md, etc.) to get current context.

user.md (personal) — ⚠️ APPEND ONLY, NEVER DELETE ⚠️

CRITICAL REMINDER: Before writing to this file, ALWAYS read existing content first to avoid duplicates and incorporate existing facts.

Raw facts and concise notes. Facts only, no polished marketing.

Example:

- Lived in Dubai 2018-2020, worked with clients across EMEA
- Lived in South Korea 2015-2017, learned Korean to intermediate level
- Built robotic arms at home using Arduino and 3D-printed parts
- Took MIT OCW Machine Learning course, applied to personal project predicting stock trends
- Won regional robotics competition in high school, team captain
- Former competitive swimmer (stopped 2016), coached youth swim team 2017-2018
- Speaks English (native), Korean (intermediate), Arabic (basic)
- Prefers remote work, available 9am-5pm EST

statements.md (personal or company) — ⚠️ APPEND ONLY, NEVER DELETE APPROVED STATEMENTS ⚠️

CRITICAL REMINDER: Once a statement is approved and written here, NEVER delete it. Only append new approved statements. Read existing statements before generating new ones to avoid duplicates.

User-approved statements only. Each entry includes:

• Statement text
• Tags
• Evidence pointer (which facts support this)

Example:

---
statement: "I bring a global perspective from living and working across three continents—from collaborating with EMEA clients in Dubai to navigating cross-cultural teams in South Korea. I've learned to adapt communication styles and build trust across cultures, which is essential for remote, distributed work."
tags: #global-living #dubai #korea #cross-cultural #remote-work #client-facing
evidence: user.md lines 1-2, 7
platform: standard cover letter
---

statement: "I'm passionate about robotics and hands-on engineering—I've built robotic arms at home using Arduino and 3D-printed custom parts. This isn't just work for me; it's how I spend my evenings."
tags: #robotics #hands-on #passion #arduino #3d-printing #side-projects
evidence: user.md line 3
platform: Upwork short pitch, technical/proof-first
---

statement: "My background in competitive swimming taught me discipline and coaching—skills I later applied when coaching a youth swim team. I understand how to break down complex skills, provide feedback, and build confidence."
tags: #coaching #leadership #discipline #teaching #past-hobbies
evidence: user.md line 6
platform: standard cover letter, leadership-focused
---

preferences.md (personal or company)

Writing style preferences by platform.

Example:

# Writing Preferences

## Tone
- Confident but not arrogant
- Warm and approachable
- Direct, minimal fluff
- Technical when relevant (show don't just tell)

## Platform Settings

### Upwork
- Length: 2-4 sentences + optional 2-3 bullet proof points
- Structure: Hook → proof → call-to-action
- Metrics-first when available

### LinkedIn / Email Outreach
- Length: 3-5 sentences
- Friendly, personal, brief
- Lead with common ground or mutual connection

### Job Portal / Formal Cover Letter
- Length: 3-4 paragraphs
- Structured: intro → experience/proof → why this role → close
- Slightly more formal tone

### Grants / Proposals
- Length: longer, evidence-heavy
- Structure: narrative with metrics and citations
- Emphasize impact and outcomes

## Avoid List
- Buzzwords: "synergy", "rockstar", "ninja", "guru"
- Overused phrases: "I'm passionate about", "proven track record"
- Sensitive info: exact salary history, medical details

## Structure Preferences
- Prefer bullets for proof points
- Use short paragraphs (3-4 sentences max)
- Open with strongest differentiator
- Close with clear next step

org.md (company)

Company/initiative facts.

Example:

# Acme Robotics Inc

## Domain
- Industrial automation, warehouse robotics
- Focus on AI-driven pick-and-place systems
- Series B funded, 50-200 employees

## Key People
- Jane Doe, Founder/CEO (former Google engineer)
- John Smith, CTO (robotics background)

## Positioning
- "We make warehouses smarter"
- Emphasis on practical, deployable solutions (not research)

## Constraints
- Remote-first company
- Looking for senior engineers with production experience

## History with User
- Applied via Upwork 2024-02-20
- Initial call scheduled for 2024-02-28

coherence.md (optional but recommended)

Track conflicts, gaps, clarifications needed.

Example:

# Coherence Notes

## Possible Conflicts
- 2024-02-24: User said "intermediate Korean" but earlier mentioned "basic conversational". Clarify level.

## Missing Info
- Need timeframe for Dubai work (years? months?)
- Need scope/metrics for robotics project (how many built? any demos?)
- Need GPA or honors for MIT OCW course mention

## Clarifications Needed
- Is Arabic "basic" = conversational or just tourist phrases?
- Does "remote work preference" mean fully remote only, or hybrid OK?

How to Use This Skill

⚠️ QUESTION CADENCE - VERY IMPORTANT ⚠️

First time with a person (onboarding):

• Ask 10 initial discovery questions in one session
• Then generate first batch of statements

All subsequent sessions (updates/refinement):

• Ask only 3 questions per round
• Then generate statements from those 3 answers
• This keeps sessions manageable and can run via cron for continuous learning

Why this matters: You can schedule this skill as a cron job to periodically learn more about the person (every week, every month, etc.). The 3-question pattern keeps token usage reasonable while continuously building your statement library.

Starting Fresh (First Run)

1. Read this skill document to understand the workflow

2. Create folder structure if it doesn't exist:

   mkdir -p skills/proposal-interview/personal/<user_name> skills/proposal-interview/companies
   

3. Create profile.md with current snapshot (name, what they do, where, objectives)

4. Explain the process to the user: "I'm going to ask you 10 onboarding questions to build your initial profile. After this first session, future updates will only ask 3 questions at a time to keep things manageable. This can even run on a schedule to continuously learn more about you."

5. Run onboarding questions (see below)

6. Ask the 10 initial discovery questions

7. Draft 3-8 statements, get approval, store approved ones

8. Future sessions: switch to 3-question rounds

Continuing Work (Existing User) — ⚠️ READ BEFORE WRITING ⚠️

CRITICAL: Always read existing files BEFORE asking questions or appending facts. This prevents duplicates and helps you incorporate existing knowledge.

1. Read existing files IN THIS ORDER:

   • skills/proposal-interview/personal/<user>/profile.md (READ FIRST - current context)
   • skills/proposal-interview/personal/<user>/user.md (facts gathered - NEVER DELETE, ONLY APPEND)
   • skills/proposal-interview/personal/<user>/statements.md (approved statements - NEVER DELETE, ONLY APPEND)
   • skills/proposal-interview/personal/<user>/preferences.md (style guide)
   • skills/proposal-interview/personal/<user>/coherence.md (gaps/conflicts)

2. Ask only 3 new questions (fill gaps, drill down on threads)

3. Append new facts to user.md (NEVER delete existing facts, always append)

4. Draft 2-4 new statements based on new facts

5. Get approval, append to statements.md (NEVER delete approved statements, always append)

Working with a Company/Initiative

1. Create company folder: skills/proposal-interview/companies/<slug>/

2. Ask company-specific questions:

   • What do they do? Domain?
   • Key people? Relationships?
   • Why is user interested?
   • What's the user's angle/fit?

3. Store company facts in org.md

4. Draft company-specific statements (e.g., "why I'm a fit for Acme Robotics")

5. Store in company's statements.md

Question Cadence (Hard Requirement)

First-ever run for a person:

• Ask exactly 10 questions before generating statements

After first batch:

• Ask 3 questions per round
• Generate statements
• Get feedback
• Update files
• Repeat

If user gives very short answers:

• Comment briefly: "That's shorter than ideal, but I can work with it."
• Ask a follow-up only if critical

Onboarding Script (Always First)

When starting with a new person, ask these setup questions:

1. Who is this for?

"Are you creating this for yourself, or for someone else (e.g., spouse, client)?"

• If for someone else, create skills/proposal-interview/personal/<their_name>/

2. Which platforms do you use?

"Which platforms do you apply through? Check all that apply:

• Upwork / Freelance marketplaces
• LinkedIn (applications or outreach)
• Email outreach (cold/warm)
• Company career portals
• Grants / funding applications
• Other (specify)"

3. Style preferences per platform

For each platform they checked, ask: "For [platform], what's your preferred style?

• Tone: confident / warm / direct / humble / technical / playful / formal
• Length: short (2-4 sentences) / medium (1-2 paragraphs) / long (3-4 paragraphs)
• Structure: bullets / narrative / metrics-first / story-first
• What does success look like? (get a reply / win a contract / advance to interview)"

4. What do you want to be hired for?

"What type of work do you want more of? What's your unfair advantage?"

5. Anything to avoid?

"Any topics, buzzwords, or sensitive info we should avoid in proposals?"

After onboarding, write preferences to skills/proposal-interview/personal/<user>/preferences.md.

The 10 Initial Questions (Template)

After onboarding, ask these 10 questions (adapt based on their work/goals):

1. Geographic & Cultural Context

"Where have you lived or traveled that shaped how you work? (Dubai, Seoul, remote teams, etc.) Answer in 3 bullets or a short paragraph."

2. Work Beyond the Resume

"What have you learned in your work that doesn't fit neatly on a resume? What was surprisingly hard and how did you handle it?"

3. What Colleagues Rely On You For

"What do colleagues, clients, or teammates repeatedly come to you for? What's your 'thing'?"

4. Education & Formative Learning

"Any classes, courses, or learning experiences that changed how you think or work? (Formal or self-taught—include MOOCs, bootcamps, workshops.)"

5. Projects & Proof

"What's a project you're proud of that shows your skills? (Side project, work project, personal tinkering, GitHub repo, portfolio piece.) Describe scope and outcome."

6. Awards & Recognition

"Any awards, scholarships, competitions, certifications, or public recognition? (School, work, industry, community.)"

7. Past Hobbies

"What hobbies did you used to do a lot but don't anymore? What skills did they build? (Sports, arts, teaching, community involvement, etc.)"

8. Books & Influences (Optional)

"Any books, ideas, or people that shaped your approach to work or leadership? (Use sparingly, only if it produces something concrete.)"

9. The Boring Story

"Tell me about a role, class, or project you found boring or frustrating. What did you learn? What do you now avoid or demand in your work?"

10. Your Unfair Advantage

"If you had to pitch yourself in one sentence, what's your unfair advantage? What makes you different from 100 other people with similar skills?"

After these 10 questions:

1. Extract facts → append to user.md
2. Draft 3-8 statements (see below)
3. Get approval
4. Store approved statements → statements.md
5. Update preferences.md based on feedback

The 3-Question Round (Subsequent Runs)

After the initial 10 questions, switch to asking 3 questions per round.

How to choose the next 3 questions:

1. Check coherence.md for gaps or conflicts that need clarification
2. Review statements.md to see which categories are under-represented
3. Drill down on strong threads:
   • If user mentioned robots → ask about tools, scope, outcomes, demos
   • If user mentioned Dubai/Korea → ask about cross-cultural teamwork, languages, client communication
   • If user mentioned a specific company → ask domain-specific questions

Question design principles:

• Allow short or long replies
• Use phrasing like: "Answer in 3 bullets or a short paragraph."
• If dates/metrics uncertain: "Approximate is fine."
• Always explain why it matters (briefly): "This helps me write more personal proposals."

After each 3-question round:

1. Extract facts → append to user.md or org.md
2. Draft 2-4 new statements
3. Get approval
4. Update files

Statement Drafting + Approval Loop

When to Draft Statements

• After the first 10 questions (initial run)
• After every 3-question round (subsequent runs)

How to Draft Statements

Generate 3-8 candidate statements per round.

Each statement must:

• Be grounded in gathered facts (no invention)
• Be usable in proposals/cover letters
• Be written in multiple variants when helpful:
  • Upwork short pitch (1-2 lines)
  • Standard cover letter (2-4 sentences)
  • Technical/proof-first variant (metrics, tools, outcomes)

Statement structure:

---
statement: "[The actual statement text]"
tags: #tag1 #tag2 #tag3
evidence: user.md line X, org.md line Y
platform: Upwork short / standard cover letter / technical
---

Approval Process

For each statement, ask the user:

"Here's a draft statement. Choose one:

1. Approve - Store as-is
2. Edit - Suggest changes (I'll rewrite and confirm)
3. Reject - Don't store, move on"

If user chooses Edit:

• Propose a rewrite
• Get confirmation
• If confirmed, store

If user chooses Approve:

• Read existing statements.md first to avoid duplicates
• Append the approved statement to statements.md in the correct folder (skills/proposal-interview/personal/ or skills/proposal-interview/companies/)
• NEVER delete previously approved statements

If user chooses Reject:

• Don't store (obviously, since it's rejected)
• Note the rejection reason in preferences.md if it reveals a style preference

⚠️ CRITICAL REMINDER: Once a statement is approved and appended to statements.md, it must NEVER be deleted. Only append new approved statements.

Update Preferences Based on Feedback

After each approval round, update preferences.md if you learn:

• Tone preferences (user prefers "confident" over "humble")
• Length preferences (user always wants shorter)
• Structure preferences (user likes bullets over narrative)
• Word choice (user hates "passionate", prefers "focused on")

Platform-Aware Statement Variants

Upwork Short Pitch (2-4 sentences + bullets)

Example:

I've built robotic arms at home using Arduino and 3D-printed parts—this isn't just work for me, it's how I spend my evenings. I bring 5 years of experience with ROS, Python, and mechatronics, plus a global perspective from working with EMEA and APAC clients.

• Built 3 functional robotic arms with custom inverse kinematics
• MIT OCW Machine Learning graduate (applied to robotics path planning)
• Available for 20-30 hrs/week, overlap with US/EU time zones

Standard Cover Letter (2-4 sentences)

Example:

I bring a global perspective from living and working across three continents—from collaborating with EMEA clients in Dubai to navigating cross-cultural teams in South Korea. I've learned to adapt communication styles and build trust across cultures, which is essential for remote, distributed work. My technical background in robotics and hands-on engineering complements my ability to work with diverse teams and deliver practical solutions.

Technical / Proof-First Variant

Example:

I've designed and built three functional robotic arms using Arduino, custom inverse kinematics algorithms (Python), and 3D-printed components. I completed MIT OCW's Machine Learning course and applied it to a path-planning optimization project that reduced movement time by 18%. My GitHub portfolio includes ROS packages, simulation environments (Gazebo), and documentation that's been forked 40+ times.

Email Outreach (3-5 sentences, friendly)

Example:

Hi [Name],

I noticed Acme Robotics is hiring for a Senior Robotics Engineer. I've been tinkering with robotic arms at home for years—building them from scratch using Arduino and 3D-printed parts—and I'd love to bring that hands-on passion to a team working on real-world automation challenges.

I've worked with clients across Dubai and South Korea, so I'm comfortable with remote collaboration and adapting to different working styles. Would you be open to a quick call to discuss the role?

Best,
[User]

Grant / Long Proposal (Evidence-heavy, structured)

Example:

My unique combination of hands-on robotics experience and cross-cultural collaboration positions me well for this initiative. Over the past four years, I have:

1. **Designed and built functional robotic systems**: I independently developed three robotic arms using Arduino microcontrollers, custom inverse kinematics algorithms (Python), and 3D-printed mechanical components. These projects demonstrate my ability to work across hardware, software, and mechanical design.

2. **Applied machine learning to robotics**: I completed MIT OpenCourseWare's Machine Learning curriculum and applied those techniques to optimize path planning for robotic arm movement, achieving an 18% reduction in task completion time.

3. **Collaborated across cultures and time zones**: My professional experience spans Dubai (EMEA clients) and South Korea (APAC teams), where I learned to navigate language barriers, cultural differences, and asynchronous communication. I speak English natively, Korean at an intermediate level, and basic Arabic.

This background enables me to contribute both technical expertise and the adaptability required for distributed, international teams.

Adaptive Question Generation

How to Decide What to Ask Next

Maintain an internal checklist of what categories are covered:

• Geographic / cultural context
• Work experience beyond resume
• Education / formative learning
• Projects / proof
• Awards / recognition
• Past hobbies
• Books / influences
• Boring stories / frustrations
• Proposal leverage / unfair advantage
• Constraints / boundaries

Prioritize questions that:

• Fill gaps in under-represented categories
• Drill down on strong signals (unique experiences like living abroad)
• Produce demonstrable proof (projects, metrics, artifacts)
• Align to target work (if a company/initiative is provided)

Triggered Recall Questions

If user mentions robots:

• "What tools did you use? What was the scope? Any metrics or demos?"
• "What's sitting in your office/workshop right now? How does it map to client value?"

If user mentions Dubai/Korea:

• "Tell me about a time cross-cultural communication was challenging. How did you adapt?"
• "What languages do you speak? At what level?"
• "How did living there change your approach to remote work or global clients?"

If user mentions a specific job/company:

• Prioritize questions that produce domain-aligned statements
• Store company-specific learnings in skills/proposal-interview/companies/<slug>/

Fact Capture Rules (Very Explicit) — ⚠️ READ THEN APPEND ⚠️

CRITICAL WORKFLOW - DO THIS EVERY TIME:

BEFORE appending any facts:

1. READ the existing user.md or org.md file to check what's already captured
2. Check for duplicates - don't re-add facts that are already there
3. Incorporate existing facts when generating statements

After each user answer:

1. Read existing facts first (see above)

2. Extract discrete factual lines from the answer

3. Append them to the correct file (NEVER delete existing content):

   • Personal facts → skills/proposal-interview/personal/<user>/user.md
   • Company facts → skills/proposal-interview/companies/<slug>/org.md

4. If a fact is uncertain, ask a one-line confirmation before writing:

   • "You mentioned living in Dubai—was that 2018-2020, or different years?"

5. Never delete or rewrite previous lines. If a conflict is detected:

   • Append the new info (DO NOT DELETE THE OLD INFO)
   • Append a "possible conflict" note to coherence.md
   • Ask a targeted clarifying question next round

REMINDER: APPEND ONLY. NEVER DELETE. ALWAYS READ BEFORE WRITING.

Example Session Flow

Session 1: Onboarding + Initial 10 Questions

[Onboarding]
Me: "Are you creating this for yourself, or for someone else?"
User: "For myself."

Me: "Which platforms do you apply through? Upwork, LinkedIn, email, job portals, grants?"
User: "Upwork and LinkedIn mostly."

Me: "For Upwork, what's your preferred style? Tone, length, structure?"
User: "Confident but not cocky, short (2-4 sentences + bullets), metrics-first."

[Preferences captured → write to preferences.md]

[Initial 10 Questions]
Me: "Question 1: Where have you lived or traveled that shaped how you work?"
User: "I lived in Dubai for 2 years (2018-2020) and Seoul for 3 years (2015-2017). Dubai taught me how to work with clients across EMEA time zones. Korea taught me patience and indirect communication styles."

[Extract facts → append to user.md]
- Lived in Dubai 2018-2020, worked with EMEA clients
- Lived in Seoul 2015-2017, learned indirect communication styles

Me: "Question 2: What have you learned that doesn't fit neatly on a resume?"
User: "I learned how to handle ambiguity. Clients don't always know what they want, and I've gotten good at asking clarifying questions early."

[Extract facts → append to user.md]
- Skilled at handling ambiguous client requirements, asks clarifying questions early

[... continue through 10 questions ...]

[After 10 questions: Draft statements]
Me: "Based on what you've shared, here are 5 draft statements. For each, tell me: Approve / Edit / Reject."

Statement 1 (Upwork short):
"I bring a global perspective from living in Dubai and Seoul, where I collaborated with EMEA and APAC clients across time zones. I'm comfortable with remote work and cross-cultural communication."
Tags: #global-living #dubai #korea #remote-work #cross-cultural

User: "Approve."

[Write to statements.md]

Statement 2 (Standard cover letter):
"I excel at navigating ambiguous client requirements. Early in projects, I ask targeted clarifying questions to uncover what clients really need—even when they don't know themselves yet."
Tags: #client-facing #ambiguity #discovery #consultative

User: "Edit—sounds too soft. Make it more confident."

Me: "How about: 'I thrive in ambiguous situations. When clients aren't sure what they need, I ask the right questions upfront to define clear goals and avoid costly rework later.'"

User: "Perfect. Approve."

[Write revised version to statements.md, update preferences.md: prefer confident tone]

[... continue through all statements ...]

Session 2: 3-Question Round

[Read existing files: user.md, statements.md, preferences.md, coherence.md]

Me: "I see you mentioned robotics earlier. Let's drill down on that."

Question 1: "You said you built robotic arms at home. How many? What tools and frameworks? Any demos or GitHub repos?"
User: "I've built 3 so far. Arduino for controllers, Python for kinematics, 3D-printed the parts. I have videos on YouTube and code on GitHub—one repo has 40+ forks."

[Extract facts → append to user.md]
- Built 3 robotic arms at home: Arduino controllers, Python kinematics, 3D-printed parts
- YouTube demos, GitHub repo with 40+ forks

Question 2: "What was the hardest part of building them?"
User: "Getting the inverse kinematics right. Took me weeks to debug the math. Eventually used a library but modified it for my joint constraints."

[Extract facts → append to user.md]
- Struggled with inverse kinematics math, debugged and modified existing library

Question 3: "How does this hobby map to the type of work you want?"
User: "I want to work on practical, deployable robotics—not research. I like solving real engineering constraints, not publishing papers."

[Extract facts → append to user.md]
- Prefers practical/deployable robotics over research, enjoys engineering constraints

[Draft new statements based on these 3 answers]
Me: "Here are 3 new draft statements based on your robotics work. Approve / Edit / Reject?"

Statement (Technical/proof-first):
"I've built 3 functional robotic arms from scratch using Arduino, custom Python kinematics, and 3D-printed components. My GitHub repo has been forked 40+ times, and I've published video demos on YouTube. I focus on practical, deployable solutions—not academic research."
Tags: #robotics #arduino #python #github #practical #hands-on

User: "Approve."

[Write to statements.md]

[... continue ...]

Safety and Integrity

No Invented Achievements

• Only write statements grounded in gathered facts
• If user didn't mention it, don't make it up

No Sensitive Data (Unless Explicit)

• Don't include salary history, medical details, or private info unless user explicitly provides and approves it

If User Requests Exaggeration or Fabrication

• Refuse politely: "I can't invent achievements, but I can help you frame what you've done more compellingly. Let's work with the real facts."
• Offer truthful alternatives

Conflicts and Uncertainty

• If something doesn't add up, note it in coherence.md and ask for clarification
• Never guess or fill in blanks

Running as a Cron Job (Continuous Learning)

You can schedule this skill to run periodically (weekly, monthly) to continuously learn more about the person and build their statement library.

Example cron job payload:

{
  "kind": "agentTurn",
  "message": "Run the proposal-interview skill for Mike. This is a continuation session (not first run), so ask 3 new questions, gather facts, draft 2-4 statements, and get approval. Focus on filling gaps in his profile or drilling deeper on OpenClaw implementation experience."
}

Benefits of cron-based learning:

• Builds statement library over time without overwhelming the user
• 3-question rounds keep sessions short and token-efficient
• Can be scheduled during low-activity periods
• Gradually captures more depth and nuance

Important: Always use the 3-question pattern for cron sessions, never the initial 10-question onboarding.

Checklist: Every Run

⚠️ CRITICAL CHECKS - READ THESE EVERY TIME:

Before you start:

• Have I identified which person/company this is for?
• Have I read profile.md FIRST to get current context?
• Have I read existing user.md or org.md to see what facts are already captured?
• Have I read existing statements.md to avoid duplicating approved statements?
• Have I read preferences.md and coherence.md?
• Have I determined if this is a first run (10 questions) or continuation (3 questions)?

During the interview:

• Am I asking questions that allow short or long replies?
• Am I reading existing facts BEFORE appending to avoid duplicates?
• Am I appending facts (NEVER deleting) to the correct file (user.md or org.md)?
• Am I noting conflicts in coherence.md instead of rewriting?
• Am I tagging statements with relevant keywords?

When drafting statements:

• Have I read existing statements to avoid duplicates?
• Are all statements grounded in gathered facts?
• Have I written variants for relevant platforms (Upwork, cover letter, technical)?
• Have I asked: Approve / Edit / Reject for each statement?
• Have I appended approved statements (NEVER deleting old ones) to statements.md?
• Have I updated preferences.md based on user feedback?

After the session:

• Have I noted any gaps or missing info in coherence.md?
• Have I identified categories that need more coverage for next round?
• Have I maintained the append-only rule (no deletions, only additions)?
• Did I update profile.md if any current information changed?

Summary

This skill transforms generic proposals into personalized, compelling letters by:

1. Interviewing you to extract unique, personal-but-professional facts
2. Drafting reusable statements grounded in those facts
3. Refining statements through an approval loop
4. Storing approved statements for future assembly

The result: a library of high-quality, pre-approved statements you can mix and match for any proposal, tailored to the platform and company.

Next steps:

• Run the onboarding + 10 initial questions
• Approve your first batch of statements
• Run 3-question rounds to fill gaps and drill down
• Build a library of 20-50 statements covering different angles
• (Future) Use those statements to assemble custom proposals on demand

File locations:

• Personal: skills/proposal-interview/personal/<name>/
• Companies: skills/proposal-interview/companies/<slug>/
• Always append, never delete
• Check coherence.md for gaps