outputai Claude Code Plugin
Claude Code PluginClaude CodeDevelopmentAI Agents & AI App DevelopmentClaude Code Customization & WorkflowThe explanation below is AI-generated. Please verify it against the sources.
outputai is a Claude Code plugin that equips the assistant to work with the Output AI development framework. It adds five specialist agents covering planning, building, debugging, prompt writing, and quality review of Output-based workflows. It also supplies more than 40 slash-command skills for tasks such as scaffolding new workflows, debugging, running evaluations, and handling credentials. A SessionStart hook loads Output SDK conventions automatically so Claude has framework context before a user's first prompt. Pricing or licensing terms for the plugin itself are unknown.
About the service
Output is an open-source TypeScript framework, described in its README, for building AI workflows and agents. It combines prompt management, evaluation, execution tracing, cost tracking, orchestration, and credential handling in a single codebase rather than relying on separate hosted tools. Prompts are kept as version-controlled files with YAML settings and Liquid templating, and workflows can call multiple LLM providers such as Anthropic, OpenAI, Azure, Vertex AI, and Bedrock through one interface. It uses Temporal internally for retries, replay, and parallel execution, and it can be run locally with Node.js and Docker via its CLI. Credentials are encrypted using AES-256-GCM and managed per environment through the same CLI.
What you can do with outputai
- Consult dedicated agents for planning, building, debugging, prompt authoring, and quality review of Output workflows
- Use slash-command skills to scaffold new Output projects and workflows
- Run and inspect evaluations for workflow quality through skill commands
- Manage encrypted credentials for Output projects via dedicated skills
- Have Claude auto-load Output SDK conventions at session start through the included hook
- Debug existing Output workflows using the provided debugging skills and agent