create

A deprecated command that creates a multi-agent workflow from a natural-language description. It remains available for older usage, but the related workflow-creation skill is preferred.

In plain words
What is it for?
Use it to describe an automated workflow, such as deploying code with security checks or implementing a feature with test-driven development, where TDD means writing tests before the code.
Why use it?
It gives existing users a way to start workflow design, while making clear that newer projects should use the replacement process.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/mbruhler/claude-orchestration/create
Clone the repo
git clone --depth 1 https://github.com/mbruhler/claude-orchestration
Per session 6 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,333 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00006 $0.01333
Opus 5 $0.00003 $0.00666
Sonnet 5 $0.00001 $0.00267
Haiku 4.5 $0.00001 $0.00133

Measured 2d ago against content hash 3c1913cae238, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

create scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/create.md · 181 lines

How it starts

The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.

⚠️ DEPRECATED: Create Workflow from Description

This command is deprecated. Use the creating-workflows skill instead for better auto-discovery and context compression.

The skill automatically activates when you describe workflows or mention automation.

Migration Guide

Instead of: /orchestration:create deploy with security validation

Just say: "Create a workflow that deploys with security validation"

The creating-workflows skill will automatically activate and guide you through workflow creation using Socratic questioning.


Legacy Usage (Still Works)

Launch the Socratic workflow designer to create workflows from natural language.

Usage

  • /orchestration:create - Start with no context, ask what to build
  • /orchestration:create <description> - Start with initial description

Examples

/orchestration:create
/orchestration:create deploy with security validation
/orchestration:create implement auth feature with TDD

Action

Parse arguments:

const description = args.trim() || null;

Launch the workflow-socratic-designer agent and handle its response:

Step 1: Call the subagent

Task({
  subagent_type: "orchestration:workflow-socratic-designer",
  description: "Create workflow from description",
  prompt: `Create an orchestration workflow from natural language description.

Initial description: ${description || "Ask user what they want to build"}

IMPORTANT: You are a subagent and do NOT have access to AskUserQuestion.
When you need to ask the user questions, return them in JSON format with "needsUserInput": true.
The main agent will use AskUserQuestion to prompt the user and call you again with answers.

Follow this process:

1. **Understand Request**
   ${description ?
     "- Assess specificity of provided description\n   - Identify workflow pattern hints" :
     "- Ask user what they want to build\n   - Gather initial context"}
   - Read existing templates for pattern matching from: ~/.claude/plugins/repos/orchestration/examples/*.flow
   - Reference agent registry for available agents: ~/.claude/plugins/repos/orchestration/agents/registry.json

2. **Socratic Questioning**
   Return questions in JSON format (you cannot use AskUserQuestion directly).

   For vague requests:
   - Problem identification (single-select)
   - Scope clarification (multi-select)
   - Constraints (multi-select)
   - Pattern suggestion (single-select)

   For specific requests:
   - Pattern recognition
   - Customization (multi-select)
   - Validation (single-select)

   For medium requests:
   - Scope first (multi-select)
   - Details drilling
   - Connection logic

3. **Build WorkflowRequirements**
   Create structured object:
   {
     intent: "user's goal",
     pattern: "identified-pattern",
     agents: ["list", "of", "agents"],
     structure: "sequential|parallel|conditional|hybrid",
     errorHandling: ["retry", "rollback"],
     checkpoints: ["@review"],
     conditions: ["if passed"],
     guards: ["require-clean-working-tree"],
     tools: ["npm:test"],
     mcps: [],
     customSyntaxNeeded: []
   }

4. **Custom Syntax (if needed)**
   If customSyntaxNeeded has elements:
   - Call workflow-syntax-designer for each
   - Use Task tool with subagent_type: "orchestration:workflow-syntax-designer"

5. **Generate Workflow**
   - Map requirements to orchestration syntax
   - Add variable bindings: operation (condition):var~>
   - Use negative conditions: (if !var)~>
   - Format for readability

6. **Explain to User**
   - Plain language explanation of workflow
   - Show generated syntax with highlighting
   - Explain any custom syntax created

7. **Save as Template**
   Return JSON to request user input for:
   - Ask if user wants to save as template
   - Collect template name (suggest based on pattern)
   - Confirm description
   - Confirm parameters
   - Save to ~/.claude/plugins/repos/orchestration/examples/<name>.flow
   - Ask which custom syntax to promote to global library
   - Copy promoted syntax to library/syntax/<type>/<name>.md

Context files:
- Templates: ~/.claude/plugins/repos/orchestration/examples/
- Global syntax: ~/.claude/plugins/repos/orchestration/library/syntax/
- Agent registry: ~/.claude/plugins/repos/orchestration/agents/registry.json
- Temp agents: ~/.claude/plugins/repos/orchestration/temp-agents/
- Natural language docs: ~/.claude/plugins/repos/orchestration/docs/features/natural-language.md
- Best practices: ~/.claude/plugins/repos/orchestration/docs/reference/best-practices.md
- Workflow syntax: ~/.claude/plugins/repos/orchestration/docs/topics/syntax.md

Remember:
- Use variable binding for explicit conditions
- Support negative conditions with !
- Follow reuse-first for custom syntax
- Make workflow self-documenting with clear variable names
`
})

Read the full file on GitHub · 181 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 181 lines · 6 tokens per session scan A 3c1913cae238

Subscribe to this mod's changes

create is a command published in the GitHub repository mbruhler/claude-orchestration (218 stars, last pushed 3mo ago), licensed MIT. It adds 6 tokens to every session and 1,333 once invoked, about $0.0000 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.