Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By wagnersza
Run multiple Claude Code instances with different auth methods from a single session.
npx claudepluginhub wagnersza/claude-multi-auth --plugin cmaCancel an active background CMA task
Show the stored final output for a finished CMA task
Manage auth profiles for multi-auth Claude Code subagents
Show active and recent CMA background tasks for this repository
Delegate a task to a Claude Code subagent using a specific auth profile
Uses power tools
Uses Bash, Write, or Edit tools
No model invocation
Executes directly as bash, bypassing the AI model
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Claude Code skill pack for Anthropic (30 skills)
Sub-agent runner — runs agent definitions on Codex, Claude Code, Cursor CLI, or Gemini CLI from any host AI tool.
Let local Claude Code sessions talk to each other in natural language.
Secret safety hooks: auto-load credentials via CLAUDE_ENV_FILE, redact secrets from CLI output. Powered by the nopeek CLI.
The most comprehensive Claude Code plugin — 14+ agents, 56+ skills, 33+ commands, and production-ready hooks for TDD, security scanning, code review, and continuous learning
Efficient skill management system with progressive discovery — 410+ production-ready skills across 33+ domains
Use Copilot from Claude Code to review code or delegate tasks.
Runs pre-commands
Contains inline bash commands via ! syntax
Runs pre-commands
Contains inline bash commands via ! syntax
Share bugs, ideas, or general feedback.
Run multiple Claude Code subagent instances from a single session, each authenticated with a different auth method (API keys, Bedrock, Vertex, LiteLLM, etc.).
This plugin is for Claude Code users who need to delegate tasks to Claude Code instances that use different authentication backends — without leaving their current session.
/cma:setup to manage named auth profiles with per-profile environment variables/cma:task to delegate work to a subagent using a specific auth profile/cma:status, /cma:result, and /cma:cancel to manage background taskscma:cma-rescue subagent for direct task delegation@anthropic-ai/claude-agent-sdk — installed automatically when Claude Code runs the pluginAdd the marketplace in Claude Code:
/plugin marketplace add wagnersza/claude-multi-auth
Install the plugin:
/plugin install cma@wagnersza
Reload plugins:
/reload-plugins
Then set up your first profile:
/cma:setup add bedrock --env CLAUDE_CODE_USE_BEDROCK=1 --env AWS_PROFILE=my-aws-profile
After install, you should see:
cma:cma-rescue subagent in /agentsOne simple first run is:
/cma:setup add bedrock --env CLAUDE_CODE_USE_BEDROCK=1 --env AWS_PROFILE=my-aws-profile
/cma:setup test bedrock
/cma:task --profile bedrock summarize the README
/cma:setupManages named auth profiles. Each profile stores environment variables that are injected into the subagent process at runtime.
Profiles are stored in ~/.claude-multi-auth/profiles.json with 0600 permissions.
/cma:setup add bedrock --env CLAUDE_CODE_USE_BEDROCK=1 --env AWS_PROFILE=my-aws-profile
/cma:setup add vertex --env CLAUDE_CODE_USE_VERTEX=1 --env CLOUD_ML_REGION=us-east5
/cma:setup add litellm --env ANTHROPIC_BASE_URL=http://localhost:4000 --env ANTHROPIC_API_KEY=sk-litellm-key
/cma:setup add direct --env ANTHROPIC_API_KEY=sk-ant-...
If you omit --env, the command will interactively ask which environment variables to set.
Common environment variables:
| Variable | Purpose |
|---|---|
ANTHROPIC_API_KEY | Direct API key authentication |
ANTHROPIC_BASE_URL | Custom API endpoint (LiteLLM, proxy) |
CLAUDE_CODE_USE_BEDROCK | Enable AWS Bedrock backend |
CLAUDE_CODE_SKIP_BEDROCK_AUTH | Skip Bedrock auth check |
AWS_PROFILE | AWS profile for Bedrock |
CLAUDE_CODE_USE_VERTEX | Enable Google Vertex backend |
CLOUD_ML_REGION | Vertex AI region |
ANTHROPIC_DEFAULT_SONNET_MODEL | Override default Sonnet model |
ANTHROPIC_DEFAULT_OPUS_MODEL | Override default Opus model |
ANTHROPIC_DEFAULT_HAIKU_MODEL | Override default Haiku model |
/cma:setup remove bedrock
/cma:setup default bedrock
/cma:setup show bedrock
Displays the profile configuration with secrets masked.
/cma:setup test bedrock
Sends a quick test message to verify the profile authenticates successfully.
/cma:setup
/cma:taskDelegates a task to a Claude Code subagent using the specified auth profile.
/cma:task --profile bedrock explain the authentication flow in this repo
/cma:task --profile vertex --model claude-sonnet-4-6 summarize the main module
/cma:task --profile litellm --write refactor the utils module
/cma:task --profile bedrock --background analyze the test coverage
/cma:task --profile bedrock --max-turns 5 find unused imports
Options:
| Flag | Description |
|---|---|
--profile <name>, -p | Auth profile to use (uses default if omitted) |
--model <model>, -m | Override model for the subagent |
--write | Allow the subagent to use write tools (Edit, Write, Bash) |
--background | Run the task in the background |
--max-turns <n> | Limit the number of agentic turns |
Note: Without
--write, the subagent runs in read-only mode. It can read files and search the codebase but cannot make changes.
/cma:statusShows running and recent tasks for the current repository.
/cma:status
/cma:status task-abc123
Use it to:
/cma:resultShows the stored final output for a finished task.
/cma:result
/cma:result task-abc123
/cma:cancelCancels an active background task.
/cma:cancel
/cma:cancel task-abc123
/cma:task --profile bedrock what does the main function do?
/cma:task --profile vertex --write add error handling to the API client