CONDUCT Wizard

CONDUCT Wizard

Building playbooks just got easier. Describe what you want your workflow to do, and Prompty designs the playbook – steps, inputs, branching, artifacts – from your description. Review it, tweak it, and you’re running in minutes instead of writing markdown from scratch.

What’s new

• Describe it, Prompty builds it – Open the CONDUCT wizard from the sidebar or dashboard, tell Prompty what your workflow should do, and she generates a complete playbook definition. The default mode is an AI consultation where you have a back-and-forth conversation about your requirements. Prefer more structure? Switch to Guided Q&A, start from a Template, or go Blank.
• @prompt expansion – Reference prompts from your library directly inside playbook steps with @prompt(name). Promptmark resolves the reference at runtime, injecting the prompt content into the step. Supports multiple schemes: library: for your library, mcp: for prompts from connected MCP servers, and bare identifiers for simple lookups.
• Dynamic prompt references – Use template variables inside @prompt directives. @prompt({{variable}}) resolves the prompt name from a runtime input, and @prompt(library:{{prompt_id}}) lets you parameterize which library prompt gets used. This is meta-prompting: workflows that select their own prompts based on context.
• Variable artifact types – Artifact output types can now reference declared inputs or captured outputs. Define type: {{output_format}} in your ARTIFACTS section, and the format adapts to whatever the workflow produces – JSON, markdown, Mermaid, or anything else.
• Improved COMPOSE quality – The COMPOSE wizard’s AI consultation produces better prompts with clearer variable design. Prompty now coaches you toward 2-5 well-defined inputs instead of vague open-ended variables, and provides concrete right/wrong examples during generation.
• Prompty usage dashboard – Track how you’re using your AI consultations across both COMPOSE (prompt creation) and CONDUCT (playbook creation), with cost estimates and usage breakdown.

The PLAYBOOK.md standard

The @prompt expansion and playbook syntax follow the PLAYBOOK.md open standard. This means playbooks you write in Promptmark are portable – the format isn’t locked to this platform. Other tools that adopt the standard can read and execute the same files.

For API integrators

The compose-playbook skill at /.well-known/agent-skills/ now supports CONDUCT-style generation. AI agents can describe a workflow and receive a valid playbook definition in response.