persona-hub-create

AI persona management for agents. Activate structured personality profiles.

Installation
CLI
npx skills add https://github.com/asimokby/persona-hub --skill persona-hub-create

Install this skill with the CLI and start using the SKILL.md workflow in your workspace.

Last updated 4/22/2026

PersonaHub

AI persona management for any agent. Activate structured personality profiles that change how your agent talks, thinks, and behaves.

Works with Claude Code, Cursor, Windsurf, Cline, GitHub Copilot, OpenClaw, Hermes, and any agent supporting the Agent Skills specification.

PersonaHub Demo — activating Trump persona in Claude Code

Install

Pick your agent. One command. Done.

Agent Install
Claude Code claude plugin marketplace add asimokby/persona-hub && claude plugin install persona-hub@persona-hub
Cursor npx skills add asimokby/persona-hub -a cursor
Windsurf npx skills add asimokby/persona-hub -a windsurf
Copilot npx skills add asimokby/persona-hub -a github-copilot
Cline npx skills add asimokby/persona-hub -a cline
Any other npx skills add asimokby/persona-hub

Install once. Use in every session after that.

Update

Use /persona-hub-update inside Claude Code, or manually:

claude plugin uninstall persona-hub@persona-hub && claude plugin install persona-hub@persona-hub

Quick Start

/persona-hub donald-trump      # Activate a persona
/persona-hub-list              # See available personas
/persona-hub-stop              # Deactivate
/persona-hub-create            # Build a new persona

Commands

Command What it does
/persona-hub <name> Activate a persona
/persona-hub-list Show available personas
/persona-hub-stop Deactivate current persona
/persona-hub-create Create or enrich a persona from folders, files, URLs, or text
/persona-hub-help Show quick-reference card
/persona-hub-update Update plugin to latest version

How It Works

Each persona is a directory of markdown files — a Personality Wiki:

personas/donald-trump/
  persona.yaml          # Manifest (name, summary, dimensions)
  identity.md           # Who they are
  voice.md              # How they talk (including anti-patterns)
  beliefs.md            # What they think (with temporal evolution)
  knowledge.md          # What they know (and don't know)
  relationships.md      # Key people
  sources/sources.yaml  # Source registry
File Agent Question Priority
identity.md "Who am I?" required
voice.md "How do I talk?" required
beliefs.md "What do I think?" recommended
knowledge.md "What do I know?" recommended
relationships.md "Who matters to me?" supplementary
biography.md "What happened to me?" supplementary

When you activate a persona, the agent reads these files and adopts the persona's voice, beliefs, and identity. Anti-patterns in voice.md are hard constraints — things the persona would never say.

Minimum viable persona = 3 files: persona.yaml + identity.md + voice.md

Key Design Principles

  • Modular — each file covers one dimension, independently updatable
  • Evolvable — files get updated as new info comes in (interviews, articles, etc.)
  • Agent-consumable — an LLM reads these files and responds in-character
  • Source-grounded — claims are traceable via inline [source:id] citations
  • Contradiction-aware — captures tensions and temporal evolution, not just facts

Creating a Persona

Run /persona-hub-create and provide sources. Accepts anything:

  • Folders — point to a directory of transcripts, interviews, or writings (best results)
  • Files — individual .txt, .md, or .json files
  • URLs — links to articles, interviews, public statements
  • Text — describe the persona in your own words

The agent analyzes all sources and builds dimension files automatically. More sources = more authentic persona.

Enriching an Existing Persona

Run /persona-hub-create again on the same persona name to add new sources. New findings are merged into existing files — nothing gets overwritten.

Or manually: copy templates/ to personas/<slug>/ and fill in the files.

Where Personas Live

Location Path Scope
Project-local ./personas/ This project only
Global ~/.persona-hub/personas/ All projects

Project-local takes precedence over global for same-slug personas.

Bundled Personas

Persona Description Dimensions Sources
Donald J. Trump 45th & 47th US President 5 (identity, voice, beliefs, knowledge, relationships) 517 transcripts, ~3.5M words

Platform Support

Platform Install Hooks Skills
Claude Code Plugin marketplace Yes Yes
Cursor Skills directory No Yes
Windsurf Skills directory No Yes
Cline Rules directory No Yes
GitHub Copilot Instructions file No Yes
OpenClaw Skills directory No Yes
Hermes Skills directory No Yes
Gemini CLI Context file No Yes

Hooks provide session persistence (remembers active persona across sessions). Without hooks, personas work within a single session.

Project Structure

persona-hub/
  skills/                           # Skill definitions (source of truth)
    persona-hub/SKILL.md            # Main skill — activate personas
    persona-hub-list/SKILL.md       # List available personas
    persona-hub-stop/SKILL.md       # Deactivate persona
    persona-hub-create/SKILL.md     # Create new persona
    persona-hub-help/SKILL.md       # Help card
    persona-hub-update/SKILL.md     # Self-update
  .claude-plugin/                   # Claude Code plugin manifest
  hooks/                            # Session hooks (Claude Code)
  commands/                         # Slash commands
  rules/                            # Auto-activation rules
  personas/                         # Bundled persona examples
  templates/                        # Starter templates
  AGENT_PROTOCOL.md                 # Full consumption specification

Full Specification

See AGENT_PROTOCOL.md for the complete specification: loading order, conflict resolution, system prompt assembly, and behavioral guidelines.

Contributing

See CONTRIBUTING.md. Edit skills/persona-hub/SKILL.md (source of truth) — CI syncs to all platforms.

License

MIT