Skip to main content

Your AI Profile

"Every session I explain who I am. How do I make that automatic?"

Your AI profile (me.md) tells every AI tool who you are, how you work, and what you already know. Instead of explaining your background at the start of every conversation, you write it once — and it travels with you.

What belongs in your profile​

Your profile captures stable identity facts:

  • Your role and domain expertise
  • Languages and tools you know well
  • Communication preferences (verbose explanations vs. terse, analogies vs. code)
  • Things the AI should never assume about you

A minimal example:

# About Me — Personal

## Background
Full-stack developer, 8 years experience.
Primary languages: TypeScript, Python, Go.

## Working Style
- Conclusions first, details on request
- Prefer explicit over implicit — spell out assumptions
- No preamble or trailing summary in responses

## Preferences
- Always show diffs for code changes, not full rewrites
- Flag breaking changes before suggesting them

What does NOT belong here​

Behavioral rules ("always run tests before pushing") belong in rules.md, not me.md. Your profile is about who you are, not how the AI should behave.

Setup​

Your profile lives at ~/.kmgraph/me.md (personal, cross-project) or knowledge/me.md (project-specific override, gitignored).

Run /kmgraph:init to create the initial profile interactively, or edit the file directly.

For cross-project personal identity (rules and style that apply across all projects), run /kmgraph:init-personal-kg. It creates ~/.kmgraph/me.md, ~/.kmgraph/rules.md, and ~/.kmgraph/triggers.md.

Project vs. personal scope​

When both files exist, the project file takes precedence. This lets you have global defaults and per-project overrides.

Each contributor keeps their own copy — it's gitignored. It covers your role on this project, your areas of strength, and what's new to you in this codebase.

# About Me — My Project

## Role
Senior backend engineer, primary maintainer of the payments service.
Strong in TypeScript and PostgreSQL; new to the React components in this repo.

## Working Style
- Direct communication: skip preamble, get to the answer
- Show diffs, not rewrites — I read code well
- Ask before refactoring anything I didn't specifically request

## What I Value
Correctness over speed. If a change might break something, say so first.

Two contributors on the same project will have different files. Committing one person's file would give every other contributor the wrong context.

Cross-platform portability​

Because me.md is plain Markdown, it works on any platform that reads KMGraph. Claude Code, Gemini CLI, Cursor — they all load the same profile. You configure it once and every AI tool benefits.

The platform shim pattern makes this work: each platform config file (CLAUDE.md, .cursorrules, GEMINI.md, etc.) becomes a thin shim that says "read knowledge/rules.md and knowledge/me.md first."

knowledge/me.md         ← your identity in this project (gitignored)
knowledge/rules.md      ← single authoritative rules source (committed)
~/.kmgraph/me.md        ← cross-project personal identity (local only)
~/.kmgraph/rules.md     ← cross-project personal rules (local only)
CLAUDE.md               ← shim: "read knowledge/rules.md and me.md"
.cursorrules            ← shim: "read knowledge/rules.md and me.md"

One profile update. Every platform served.

Model tier configuration​

me.md also accepts an optional platforms[] frontmatter block to override model tier mappings for this project. See Model Tier Resolver for the full schema.