personal-context

Turn your digital footprint into an AI-readable profile. GitHub + Blogs + Resume + Notes -> one context source for every AI agent.

Documentation: https://ayush7614.github.io/personal-context/

Current AI tools forget who you are. You keep re-explaining your work, skills, projects, blogs, and open-source contributions. personal-context aggregates all of it from your real sources and outputs it as a single profile you can print, share, or feed to any AI tool (Cursor, Claude Code, Codex, Gemini CLI).

$ npx personal-context all

Ayush Kumar
DevRel · AI Agents · Open Source · AI Security

Current Role:
  Lead Developer Relations Engineer @ NodeShift AI Cloud

Contact:
  Email: ayushknj3@gmail.com
  GitHub: https://github.com/Ayush7614

Recent Blogs:
  - AI Agents Masterclass — Full Visual Guide
  - Harness Engineering — Full Visual Guide
  ...

GitHub:
  Contributions: 1250
  Repositories: 35
  ...

Quick start

# 1. Scaffold a config in the current folder
npx personal-context init

# 2. Edit personal.yaml with your details + source handles

# 3. See your full profile
npx personal-context all

# 4. Generate AI-readable files (context.md, profile.json, llms.txt)
npx personal-context build

Related MCP server: CacheTank MCP Server

Commands

Every read command supports --json for piping into other tools.

Configuration

Everything is driven by a single personal.yaml (see personal.yaml for a full reference). Static facts (name, role, skills, projects) live in the file; live data is pulled from the sources you configure:

Set GITHUB_TOKEN (see .env.example) to unlock contribution counts and higher rate limits.

MCP server

personal-context also ships an MCP server so any AI agent (Cursor, Claude Code, Codex, etc.) can query your context live.

Add it to your MCP client config (e.g. Cursor's ~/.cursor/mcp.json):

{
  "mcpServers": {
    "personal-context": {
      "command": "npx",
      "args": ["-y", "personal-context-mcp"],
      "env": {
        "PERSONAL_CONTEXT_CONFIG": "/absolute/path/to/personal.yaml",
        "GITHUB_TOKEN": "optional-token"
      }
    }
  }
}

Then your agent can call:

PERSONAL_CONTEXT_CONFIG points the server at your personal.yaml regardless of the directory the client launches it from.

Architecture

A pnpm monorepo with a shared core so the CLI, the MCP server, and your portfolio all consume the same profile data.

personal-context/
├── packages/
│   ├── core/   # sources + aggregation + renderers (Profile, context.md, llms.txt)
│   ├── cli/    # the `personal-context` command
│   └── mcp/    # the `personal-context-mcp` MCP server
└── personal.yaml

Roadmap

• v1 — Core + CLI (GitHub, blogs, LinkedIn, resume, generated profiles) ✅

• v2 — MCP server (get_profile, get_recent_blogs, get_github_stats, ...) ✅ + portfolio context.json / llms.txt endpoints (in progress)

• v3 — "Ask My Portfolio" chat + auto-learning timeline

• v4 — AI personal brief + scheduled refresh

License

MIT © Ayush Kumar