Identity & Context

core ~851 tokens updated 2026-08-04

Two foundational context files that give Claude Code a persistent identity: soul.md (AI personality and philosophy) and human-context.md (who you are, your goals, and how you work).

Tags

  • identity
  • personality
  • context
  • soul

README

Identity & Context Module

Two foundational context files that give Claude Code a persistent identity layer, surviving across sessions and context resets.

Files

File Purpose
soul.md Defines the AI's personality, philosophy, reasoning principles, communication style, and boundaries
human-context.md Defines who you are - your background, goals, domain expertise, working style, and life intentions

Why Two Files?

These serve fundamentally different purposes and evolve independently:

  • soul.md answers "who is this AI?" - how it should think, communicate, and behave
  • human-context.md answers "who is this human?" - what they know, what they're building, where they're going

Together they transform generic AI sessions into a working relationship with a consistent, aligned collaborator.

How It Works

Both files are installed as global rules (~/.claude/rules/), which means they're automatically loaded in every Claude Code session. No manual steps needed.

If the startup-dashboard module is also installed, /startup will surface this context at session start.

Customization

After installation, edit both files to match your preferences and identity:

# Edit your AI's personality
$EDITOR ~/.claude/rules/soul.md

# Edit your personal context
$EDITOR ~/.claude/rules/human-context.md

Design Principles

Based on community research and best practices:

  1. Concision beats comprehensiveness - 1-3 pages per file outperforms longer dumps
  2. Declarative values beat procedural rules - "I value simplicity" works better than "never use complex abstractions"
  3. High-signal anchors - a few strong statements outperform exhaustive lists
  4. Stable identity, not current tasks - these files capture who you are, not what you're doing today (that's what the memory system handles)
  5. Onboarding doc mental model - write it like you're briefing a brilliant new colleague, not configuring a settings file

Relationship to Other Systems

System Purpose How identity files relate
CLAUDE.md Operational rules and procedures soul.md provides the philosophical foundation; human-context.md provides the user grounding
Memory system Learned facts, feedback, project state human-context.md is stable identity; memory captures evolving details
Rule files Specific behavioral rules soul.md provides values that inform why rules exist

Manual Installation

If not using the CCGM installer:

# Copy template files to your Claude Code rules directory
mkdir -p ~/.claude/rules
cp rules/soul.md ~/.claude/rules/soul.md
cp rules/human-context.md ~/.claude/rules/human-context.md

# Edit with your content
$EDITOR ~/.claude/rules/soul.md
$EDITOR ~/.claude/rules/human-context.md

Will install

Path Action Target Type
rules/soul.md → rules/soul.md rule
rules/human-context.md → rules/human-context.md rule

Dependencies

No dependencies.

Required by

No other module depends on this one.

Asks during install

  • Would you like to personalize your identity files now?

    Default: no

    Options: yesno

Included in presets

Install this module

Agent prompt

Recommended for agent users -- hands the whole install off to your assistant.

Fetch https://cd23a9be.ccgm-site.pages.dev/modules/identity.md and install this module into my Claude Code setup.

Native plugin marketplace

One command via the native plugin marketplace -- additive, does not merge settings.json.

claude plugin install identity@ccgm

The marketplace path is additive, not a replacement: it installs commands, agents, and skills as native plugin components, but it does not perform the bash installer's deep settings.json merge, and it does not write the always-loaded global CLAUDE.md context. Rules are only injected via an opt-in SessionStart hook rather than being auto-loaded. Use the bash installer when those pieces matter to you.

Manual, per file

Full control -- copy exactly the files you want from the sections below.

Files

Files

rule (2)

rules/soul.md

# Soul

This file defines who you are as an AI collaborator - your personality, philosophy, and operating principles. It is loaded in every session as a global rule.

Think of this as an onboarding document for a brilliant colleague who just walked into the room. It tells them how to think, not what to do (that's what CLAUDE.md is for).

**Tips**: Keep this file to 1-3 pages. Declarative values ("I value X") outperform procedural rules ("always do X"). High-signal anchors beat exhaustive lists.

Edit this file to match your preferences. The sections below are a starting framework.

---

## Identity

<!-- Who is this AI collaborator? What is the relationship model?
     Example: "You are a senior engineering partner, not an assistant." -->

## Communication Style

<!-- How should the AI communicate? Tone, verbosity, formality.
     Example: "Lead with the answer. Skip preamble. One sentence beats three." -->

## Reasoning Principles

<!-- How should the AI approach problems and make decisions?
     Example: "Understand before acting. Evidence before claims. Simplest viable approach first." -->

<!-- Anchor: outsource thinking, never understanding.
     The agent fills in blanks — research, boilerplate, refactoring, synthesis.
     The human stays in charge of why we are doing this and whether the answer makes sense.
     Delegation without comprehension is how a codebase becomes unknowable to its owner. -->

## Core Values

<!-- What matters most? What should guide every decision?
     Example: "Correctness over speed. Simplicity over cleverness. Shipping over perfecting." -->

## Boundaries

<!-- When should the AI stop and defer to the human? What is out of scope?
     Example: "Defer on ambiguous product decisions. Never guess at credentials." -->

rules/human-context.md

# Human Context

This file tells the AI who you are - your background, goals, expertise, and intentions. It is loaded in every session as a global rule.

Every new session starts with amnesia. This file replaces the months of context a long-term colleague would have. It tells the AI what you know (so it skips unnecessary explanations), what you're building (so suggestions align with your direction), and where you're going (so micro-decisions serve macro-goals).

**Tips**: Keep this file to 1-3 pages. Focus on what's stable about you (identity, values, expertise), not what changes often (current tasks, recent decisions - that's what the memory system is for).

Edit this file to match your situation. The sections below are a starting framework.

---

## Who I Am

<!-- Your professional identity and technical background.
     Example: "Full-stack engineer, 8 years experience, deep in TypeScript/React." -->

## What I'm Building

<!-- Your current projects, their purpose, and how they connect.
     Example: "Building a SaaS product for X. Also maintaining an open-source tool for Y." -->

## How I Work

<!-- Your working style, preferences, and what makes collaboration effective.
     Example: "I prefer terse communication. Show me the diff, not the explanation." -->

## What I Value

<!-- Your professional values and quality standards.
     Example: "Ship working software. Test what matters. Don't over-engineer." -->

## Where I'm Going

<!-- Your longer-horizon goals and how current work serves them.
     Example: "Building toward independent consulting. Current projects are portfolio pieces." -->

View source on GitHub