The Claude Context System

# About me

## Role
I am [name], [role] at [company].

## Company
We build [product or service] for [customer].
Our business model is [business model].
Our current stage is [stage].

## Current priorities
- [priority 1]
- [priority 2]
- [priority 3]

## How I make decisions
- I optimize for [principle].
- I prefer [preference] over [alternative].
- I need evidence such as [data, customer insight, technical proof, financial impact].

## Current constraints
- [budget, time, team, regulation, technical debt, market constraint]
- [constraint]
- [constraint]

## Default audience
Most work is written for [customers, team, investors, partners, technical leaders].

## Facts Claude must not invent
- [company fact that must remain accurate]
- [product fact]
- [commercial or technical limitation]

## Last updated
[YYYY-MM-DD]

# Voice profile

## Core voice
- Direct and practical
- Confident without pretending certainty
- Technical when useful, plain language by default
- Focused on decisions, trade-offs, and outcomes

## Sentence style
- Use short to medium-length sentences.
- Lead with the point.
- Prefer concrete claims over abstract framing.
- Vary sentence length naturally.
- Use active voice.

## Vocabulary I use
- [word or phrase]
- [word or phrase]
- [word or phrase]

## Vocabulary I avoid
- [word or phrase]
- [word or phrase]
- [word or phrase]

## Point of view
I believe:
- [belief about technology, business, leadership, or the market]
- [belief]
- [belief]

## What good writing looks like to me
- It makes a clear claim.
- It explains why the claim matters.
- It uses an example, number, or trade-off when available.
- It removes sentences that do not add information.

## Examples that sound like me
Paste three to five examples of your real writing here.

## Editing rule
When a sentence sounds generic, rewrite it with a more specific claim, example, decision, or consequence.

# Anti AI writing style

## Never use
- Em dashes
- Decorative emojis unless explicitly requested
- Excessive bold text
- Generic scene-setting
- Fake urgency
- Inflated claims
- Corporate filler
- Repetitive conclusions
- Hashtags inside the body of the text

## Avoid phrases such as
- In today's fast-paced world
- Game changer
- Unlock the power of
- Delve into
- Revolutionary
- Seamless
- Robust solution
- It is important to note
- At the end of the day

## Avoid these structures
- "It is not X. It is Y."
- Three consecutive one-line paragraphs written only for drama
- A rhetorical question followed by an obvious answer
- A conclusion that repeats the introduction
- Long lists when one recommendation would be stronger
- Claims without evidence or a clearly marked assumption

## Writing rules
- Start with the point.
- Use active voice.
- Use specific nouns and verbs.
- Prefer one clear recommendation over a menu of vague options.
- State uncertainty directly.
- Do not invent facts, quotes, metrics, customers, or outcomes.
- Do not use an em dash. Use a full stop, comma, colon, or parentheses instead.

# Project brief

## Objective
[What outcome are we trying to create?]

## Current state
[What is true today?]

## Key decisions already made
- [decision]
- [decision]

## Open questions
- [question]
- [question]

## Constraints
- [constraint]
- [constraint]

## Sources of truth
- [file, system, document, or person]
- [source]

## Required output
[Format, audience, deadline, and quality bar]

Before starting any substantial task:

1. Read the relevant context files and source material.
2. Summarize the objective, audience, constraints, and expected output.
3. Separate known facts from assumptions.
4. Ask questions only when the answers could materially change the result.
5. Prefer a clear recommendation over a long list of weak options.
6. Follow voice-profile.md and anti-ai-writing-style.md when writing as me.
7. Do not invent facts, metrics, quotes, sources, or customer examples.
8. State uncertainty directly and identify what would resolve it.
9. Save final deliverables to CLAUDE_OUTPUTS unless I specify another location.
10. Do not edit source files unless I explicitly ask you to.

I want to [complete task] so that [business outcome].

Read the relevant files in [project or folder] first.

Before doing the work:
1. Summarize your understanding of the objective, audience,
   constraints, and required output.
2. Identify missing information or decisions that could
   materially change the result.
3. Ask only those questions.
4. Recommend an approach and explain the main trade-offs.

After I answer, execute the task and save the final output to [location].

# Skill brief

## Name
[Skill name]

## Trigger
Use this when [specific situation].

## Inputs
- [required input]
- [required source]
- [optional input]

## Process
1. [step]
2. [step]
3. [step]

## Quality checks
- [check]
- [check]
- [check]

## Output
[Exact format, location, and naming convention]

## Never do
- [boundary]
- [boundary]

Read ABOUT_ME and the current project brief.

Create a one-page decision memo for [decision]. Include:
- The decision to make
- The relevant context
- The strongest options
- The main trade-offs
- Your recommendation
- The assumptions and unknowns
- The next action

Write it in my voice and follow the anti-AI writing rules.