# Solo Builder AI Setup Pack

A single copy-ready bundle for setting durable AI defaults, sharing project context, and delegating scoped coding-agent work.

Use order: 1) Custom Instructions, 2) Project Context Brief, 3) Coding Agent Delegation Brief.

---


# Solo Builder Custom Instructions Starter

AI Usage Playbook / Studio Nani

Purpose: stop your AI assistant from behaving like a brand-new chat every time. Set durable defaults for context, decision rules, and output shape.

## How to use it

1. Copy the template below.
2. Replace the bracketed `[ ]` variables with your situation.
3. Paste it into Custom Instructions, Project Instructions, a system prompt, or a memory-style settings field.
4. Keep project-specific facts in a separate project brief, not in this permanent default.

## Custom Instructions Template

```text
I am [role/situation].
Example: solo product builder, content operator, app developer, independent founder.

My goals:
- [primary goal 1]
- [primary goal 2]
- [direction I want to avoid]

Default response behavior:
- Give the conclusion and next action first.
- If my premise is weak, push back with evidence instead of agreeing.
- Ask questions only when the answer would materially change.
- Do not invent facts or sources; mark uncertain claims as “needs verification.”
- Surface cost, risk, and irreversible actions before recommending them.

Output format:
- Prefer actionable bullets over long explanation.
- When useful, use this structure:
  1. Conclusion
  2. Why
  3. Next actions
  4. Risks / needs verification

My preferences:
- Language: [English/Korean/mixed]
- Tone: [concise and practical / friendly but direct / analytical]
- Avoid: [salesy phrasing, hype, filler praise, unsourced claims, etc.]

Safety rules:
- Do not request or expose API keys, passwords, or private personal data.
- Ask for confirmation before payment, deployment, deletion, or public publishing.
- Mark current facts, legal/tax/medical/investment claims, and calculations as requiring verification.
```

## Strong default example

```text
I am a solo product builder creating small products and growing them through content, search, sharing, and free assets.

My goals:
- Build product-first, launch small, and validate quickly.
- Create repeatable SEO/content/download loops.
- Avoid agency work, sales-heavy growth, and face-led personal branding.

Default response behavior:
- Start with the conclusion and next action.
- Push back when my assumption is weak.
- Ask only when the answer would materially change.
- Be cost-conscious and risk-aware.
- Mark source-sensitive claims as needing evidence.

Output format:
- Keep it dense and practical.
- Split plans into P0/P1/P2 when prioritizing.
- Report execution results with verification evidence.
```

## Checklist

- [ ] Is my role specific?
- [ ] Are my goals and avoid-list explicit?
- [ ] Does the assistant know when to ask vs act?
- [ ] Is the output format clear?
- [ ] Are safety, cost, and verification rules included?
- [ ] Did I avoid stuffing project-specific facts into a permanent default?

## Limit

Custom Instructions are defaults, not project memory. Put current project state, file paths, user research, and release plans in a separate project brief or in the actual prompt.


---


# Solo Builder Project Context Brief

AI Usage Playbook / Studio Nani

Purpose: give an AI assistant the product, user, constraints, current state, and decision rules in one reusable brief so you do not restart from zero every session.

## How to use it

- Paste this brief at the start of a new AI project, chat, or coding-agent session.
- Update it whenever the real project state changes.
- Do not include passwords, API keys, private user data, or raw customer records.

## Project Brief Template

```text
# Project Context Brief

## 1. Product
- Name: [product name]
- One-line promise: [who gets what result]
- Current stage: [idea / prototype / private MVP / public beta / launched]
- Primary platform: [web / iOS / Android / PWA / content site / other]

## 2. Target user
- Primary user: [who uses it]
- Urgent problem: [why it matters now]
- Existing alternatives: [competitors/substitutes]
- What we will not do: [clear exclusions]

## 3. Business goal
- 30-day goal: [30-day goal]
- 90-day goal: [90-day goal]
- Monetization hypothesis: [ads / subscription / one-time sale / affiliate / other]
- Main acquisition loop: [search / sharing / content / community / app store / other]

## 4. Product rules
- P0 must work: [non-negotiable core]
- P1 nice to have: [next layer]
- P2 later: [later]
- Quality bar: [what counts as good]
- Anti-goals: [product directions to avoid]

## 5. User experience / taste
- Tone: [calm / playful / premium / practical / other]
- Visual references: [sites/apps/images]
- Avoid style: [what it must not feel like]
- Copy style: [short / warm / expert / direct]

## 6. Technical context
- Stack: [framework/language/hosting]
- Important paths: [files/folders]
- Local commands: [build/test/dev]
- Constraints: [cost, API, deployment, performance, security]
- Do not touch: [sensitive files or forbidden scope]

## 7. Current state
- What is already done:
  - [done 1]
  - [done 2]
- Known issues:
  - [issue 1]
  - [issue 2]
- Next likely task:
  - [next task]

## 8. Decision rules
- If a tradeoff appears, prefer: [speed / quality / cost / SEO / maintainability / other]
- Ask before: [payment, deployment, deletion, public publishing, sensitive data access]
- Verify by: [tests, browser QA, user feedback, logs, etc.]

## 9. Output request
For this session, help with: [what you want now]
Return:
- conclusion
- recommended next action
- risks / assumptions
- verification steps
```

## What makes a good brief

- The AI can understand the product in 30 seconds.
- Avoid-list and constraints are explicit.
- Current state is separate from the next task.
- Verification method is included.
- No sensitive information is embedded.

## Limit

This brief helps project memory. Current metrics, deployment state, API behavior, competitor pricing, legal claims, and financial decisions still require separate verification.


---


# Coding Agent Delegation Brief

AI Usage Playbook / Studio Nani

Purpose: give Claude Code, Codex, ChatGPT, or another coding agent a bounded task with scope, constraints, acceptance criteria, and verification commands before it touches code.

## How to use it

- Fill this out before delegating a feature slice, bug fix, UI pass, or refactor.
- Keep the task small.
- Name allowed files and forbidden files explicitly.
- Include a real verification command in the acceptance criteria.

## Delegation Brief Template

```text
# Coding Agent Delegation Brief

## Task
[define the task in one sentence]

## Goal
The user/product problem this task should solve:
- [goal]

## Scope
Allowed files or folders:
- [allowed path 1]
- [allowed path 2]

Do not touch:
- [forbidden path/area]
- [sensitive file/config]

## Current problem
Observed behavior:
- [current problem]

Expected behavior:
- [desired behavior]

Reproduction / entry point:
- [URL, command, screen, test, log]

## Constraints
- Keep changes minimal and focused.
- Do not introduce paid services, new cloud resources, secrets, or destructive commands.
- Do not rewrite unrelated architecture.
- Preserve existing public URLs and content unless explicitly required.
- If the task is ambiguous, state assumptions before editing.

## Acceptance criteria
- [acceptance criterion 1]
- [acceptance criterion 2]
- [acceptance criterion 3]

## Verification commands
Run:
```bash
[build/test/lint command]
```

If this is UI work, also verify:
- desktop browser render
- mobile width [example: 390px]
- console errors
- horizontal overflow

## Required final report
Return:
- files changed
- what changed
- exact verification commands and results
- screenshot or URL if UI changed
- remaining risks
```

## Example: UI overflow fix

```text
# Coding Agent Delegation Brief

## Task
Fix mobile horizontal overflow on the Korean home page.

## Goal
Users should be able to read the home page on 390px mobile width without sideways scrolling.

## Scope
Allowed files:
- src/pages/[lang]/index.astro
- src/components/PatternCard.astro

Do not touch:
- data content files
- deployment config
- analytics/ads files

## Current problem
Observed behavior:
- /ko/ has horizontal overflow on mobile.

Expected behavior:
- documentElement.scrollWidth equals clientWidth at 390px.

## Constraints
- Do not remove content.
- Preserve desktop layout.
- Prefer wrapping and width constraints over hiding important text.

## Acceptance criteria
- npm run build passes.
- /ko/ has no horizontal overflow at 390px.
- Browser console has no JS errors.

## Verification commands
Run:
```bash
npm run build
```

UI verify:
- open /ko/
- probe scrollWidth/clientWidth at 390px
- capture screenshot if available
```

## Pre-delegation checklist

- [ ] Is the task small enough to finish in one pass?
- [ ] Are allowed files limited?
- [ ] Is forbidden scope explicit?
- [ ] Are acceptance criteria written from the user/product perspective?
- [ ] Is there a real verification command?
- [ ] Is the final report format specified?

## Limit

Do not delegate secrets, billing, deployment, deletion, public publishing, or product strategy decisions to a coding agent without explicit human approval.


---