Multi-persona layer for Claude Code. Per chat (or per CLI call) a different agent role: coder (default), browser, inbox, research, ... Each persona has its own tools, MCP servers, system prompt and workspace. Optional, on top of voice or standalone. Trigger when the user mentions cowork, persona, agent role, multi-agent, bind chat, /cowork-*.
Scanned 9/3/2026
Install to Claude Code
npx -y skills add CorvinLabs/CorvinOS --skill cowork --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Cowork?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/corvinlabs-cowork)More formats (shields.io, HTML) on the badges page.
---
name: cowork
description: Multi-persona layer for Claude Code. Per chat (or per CLI call) a different agent role: coder (default), browser, inbox, research, ... Each persona has its own tools, MCP servers, system prompt and workspace. Optional, on top of voice or standalone. Trigger when the user mentions cowork, persona, agent role, multi-agent, bind chat, /cowork-*.
---
# cowork — personas for Claude Code
cowork is an **optional plugin on top of Claude Code** (and integrates
automatically with voice/bridges if both are installed). It turns a single
coding agent into a pool of roles that switch depending on the chat or CLI
call.
## Mental model
```
chat_profiles[<chat>].persona = "browser"
│
▼
cowork.resolver.resolve("browser", overrides=chat_profile)
│
▼
{cwd, mcp_servers, tools, append_system, permission_mode}
│
▼
claude --mcp-config ... --append-system-prompt ... --add-dir ...
```
## What you, as Claude, need to know
- **Personas live in JSON files** under `operator/cowork/personas/` (bundle)
or `~/.config/claude-cowork/personas/` (user; shadows the bundle).
- **Schema:** `name`, `description`, `permission_mode`, `allowed_tools`,
`disallowed_tools`, `append_system`, `model`, `mcp_servers` (dict in
Claude Code MCP config format), `add_dirs` (list), `needs_keys`,
`needs_oauth`, `zero_config` (bool).
- **Voice integration is additive:** `chat_profiles[<chat>].persona = "..."`
is a new optional field next to `permission_mode` / `allowed_tools` / etc.
Profile fields merge on top of the persona (lists union, scalars
overwrite, `append_system` concatenates).
- **Hot-reload remains:** the persona file is read fresh per inbox message,
no restart needed.
- **Standalone:** `cowork run <persona> -p "..."` runs `claude` locally
without involving voice/bridges.
## When the user asks ...
- "which personas do I have" → `/cowork-list`
- "what can persona X do" → `/cowork-show X`
- "make this chat the browser persona" → `/cowork-bind <channel> <chat> browser`
- "undo that" → `/cowork-unbind <channel> <chat>`
- "I want to give inbox its own MCP servers" → `/cowork-add inbox` (copies
it into the user dir, then edit)
- "test browser persona without the bridge" → `/cowork-run browser -p "..."`
## What NOT to do
- **Never** set `chat_profiles.default.persona` without an explicit request —
that would reroute every chat (same footgun rule as for
`chat_profiles.default` in the voice plugin).
- **Never** claim a persona "is running" when `mcp_servers` is empty and
`needs_oauth` / `needs_keys` are not satisfied. Give a setup hint instead.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!