Fluent Commerce Logo
Docs

Workflow and Manifest Tools in Fluent CLI

Essential knowledge

Intended Audience:

Technical User

Author:

Ben Harrison

Changed on:

17 June 2026

Overview

Fluent workflows and Mystique manifests are often large, complex JSON files that are difficult for both humans and AI assistants to work with. The Fluent CLI provides a set of commands to summarize, extract, and merge these files, enabling an efficient development workflow.

Key points

  • Large Files: Production workflows can contain hundreds of rulesets; manifests can have dozens of pages
  • Complex Structure: Nested dependencies, event chains, and component hierarchies are hard to navigate
  • AI Context Limits: LLMs struggle to parse and reason about 10,000+ line JSON files
  • Version Control: Diffing large JSON files is ineffective for code review

Why These Tools?

The CLI provides three complementary operations:
  • Summarize: Generate a high-level view of structure and dependencies
  • Extract: Pull out a focused subset for modification
  • Merge: Integrate changes back into the source
This enables a workflow where you analyze the whole, extract the part you need, modify it, and merge it back.

Workflows

Workflows define the state machine for Fluent entities (ORDER, FULFILMENT, etc.) using the Rubix orchestration engine.

Workflow Summarize

Analyze workflow structure and display ruleset dependency chains.Output: A hierarchical tree showing how rulesets connect through event production chains.The structure of each ruleset is as follows:
  • ref: Each ruleset is assigned a reference ID (e.g., "creat", "booke") for use with `extract`.
  • type_code: A shorthand representing the entity type, e.g. "Or" for Order, "Fu" for Fulfilment
  • ruleset_name and trigger_states are as per workflow definitions
Example output:Use cases:
  • Understand complex workflow dependencies
  • Identify potential loops or orphaned rulesets
  • Find which rulesets produce specific events
  • Get reference IDs for extraction

Workflow Extract

Extract a subset of a workflow containing a specific ruleset and all its descendants.Parameters:
  • `--ref`: Reference ID from `workflow summarize` output (e.g., "creat", "booke")
  • `-o`: Output file path (optional, defaults to stdout)
Output: A partial workflow JSON containing:
  • The specified ruleset and all descendants
  • Required statuses referenced by extracted rulesets
  • Metadata in `_partial` field tracking extraction details
Use cases:
  • Focus on a specific workflow branch for analysis
  • Create smaller files for AI assistants to process
  • Extract a subtree for development, testing, or documentation
  • Isolate changes for code review

Workflow Merge

Merge a workflow fragment into a target workflow, with token replacement.Parameters:
  • `<fragment>`: Path to workflow fragment file
  • `<targetworkflow>`: Path to local workflow file OR workflow name (requires profile/retailer)
Process:
  • Loads fragment and target workflow
  • Resolves template tokens (e.g., `{{RETAILER_NAME}}`)
  • Validates entityType and entitySubtype match
  • Merges rulesets, statuses, and user actions by key
  • Writes merged result to `<targetworkflow>.merged.json`
  • Uploads merged workflow to API
Merge behavior:
  • Rulesets matched by `name`
  • Statuses matched by `name`
  • User actions matched by `userActionKey`
  • Triggers matched by `trigger.key` (preferred) or `trigger.name` (legacy)
  • New items are added, existing items are replaced
Token replacement:
  • Tokens like `{{RETAILER_NAME}}`, `{{MODULE_NAME}}` are replaced using CLI config
  • Template values loaded from module config and profile settings
Use cases:
  • Install workflow fragments from modules
  • Apply customizations to base workflows
  • Merge changes from extracted and modified fragments

Manifests

Mystique manifests define the structure of Fluent web applications: navigation, pages, sections, and components.

Manifest Summarize

Analyze manifest structure and display sections, pages, and component hierarchy.Output: A structured view of sections, pages, and components. When loading from API, all referenced fragments are downloaded to the output directory.Example output:Use cases:
  • Understand manifest structure across fragments
  • Find page paths for extraction
  • Map component usage across pages
  • Audit manifest completeness

Manifest Extract

Extract a specific page from a manifest by path, wrapped in metadata envelope for later merging.Parameters:
  • `--path`: Page path to extract (e.g., "orders", "orders/:id")
  • `-o`: Output file path (optional, defaults to stdout)
Output: A manifest extract envelope with metadata:Use cases:
  • Extract a single page for focused modification
  • Create smaller files for version control
  • Isolate page changes for code review
  • Share page definitions across manifests

Manifest Merge

Merge a manifest extract back into a target manifest using action-based strategies.Parameters:
  • `<extract>`: Path to extract file (must contain `_extract` envelope)
  • `<targetmanifest>`: Path to local manifest file OR manifest name (requires profile)
  • `-o`: Output file path (default: `{fragmentName}.merged.json`)
  • `--upload`: Upload merged fragment to API as setting
Merge actions (specified in extract `_extract.action`):
  • `REPLACE`: Replace existing page at path
  • `ADD`: Add new page (error if exists)
  • `DELETE`: Remove page at path
Process:
  • Loads extract envelope and target manifest
  • Identifies which fragment contains the target page
  • Applies action (REPLACE/ADD/DELETE) to fragment
  • Writes merged fragment to output file
  • Optionally uploads fragment to API as setting
Use cases:
  • Apply extracted page modifications back to source
  • Add new pages to existing manifests
  • Remove deprecated pages
  • Deploy manifest changes to environments

Development Workflow

Recommended Pattern: Analyze → Extract → Modify → Merge → Test

For Workflows
For Manifests

Benefits for AI Assistants

When working with Claude Code or other AI assistants:
  • Summarize first: Let the AI see structure without processing the entire file
  • Extract focused subset: AI can process a 200-line extract instead of a 5,000-line file
  • Modify extracted file: AI stays within context limits while making precise changes
  • Merge with confidence: Automated merge ensures correct integration
Example prompt to Claude:Claude can now work efficiently, using the CLI tools to stay within its context window.