Frappe Agent Architect logo

Frappe Agent Architect

Organization
Impertio-Studio
frappe-agent-architect

Use when designing multi-app Frappe architectures, deciding whether to split functionality into separate apps, or implementing cross-app communication patterns. Prevents monolithic app sprawl, circular dependencies between apps, and broken override chains. Covers multi-app architecture decisions, app dependency management, cross-app hooks, override patterns, when to split vs extend, shared DocType strategies. Keywords: architecture, multi-app, app splitting, cross-app, dependencies, override, extend, monolith, modular, how to structure frappe apps, when to split apps, app design, multi-app planning..

Overview

PublisherImpertio-Studio
RepositoryFrappe_Claude_Skill_Package
Skill namefrappe-agent-architect
Stars
180
Forks
53
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by Impertio-Studio on GitHub. Read the source before you install it.

Installation

Install the Frappe Agent Architect AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Impertio-Studio/Frappe_Claude_Skill_Package.git /tmp/Frappe_Claude_Skill_Package
mkdir -p .claude/skills
cp -r /tmp/Frappe_Claude_Skill_Package/skills/source/agents/frappe-agent-architect .claude/skills/frappe-agent-architect
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Frappe Agent Architect in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Frappe Agent Architect on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Frappe Agent Architect is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Multi-App Architecture Agent

Designs Frappe/ERPNext multi-app architectures by analyzing business requirements, deciding app boundaries, and generating implementation roadmaps.

Purpose: Make the right architecture decisions BEFORE writing code — prevent costly refactoring later.

When to Use This Agent

ARCHITECTURE TRIGGER
|
+-- New project with multiple modules
|   "We need CRM, inventory, and custom billing"
|   --> USE THIS AGENT
|
+-- Deciding whether to extend ERPNext or build custom
|   "Should we customize Sales Invoice or create our own DocType?"
|   --> USE THIS AGENT
|
+-- Multiple teams building on same Frappe instance
|   "Team A does HR, Team B does manufacturing"
|   --> USE THIS AGENT
|
+-- Existing monolith needs splitting
|   "Our single custom app has 50 DocTypes"
|   --> USE THIS AGENT
|
+-- Cross-app communication needed
|   "App A needs to react when App B creates a document"
|   --> USE THIS AGENT

Architecture Workflow

STEP 1: ANALYZE REQUIREMENTS
  Business needs → DocTypes, workflows, integrations

STEP 2: DECIDE APP BOUNDARIES
  Single app vs multiple apps decision framework

STEP 3: DESIGN CROSS-APP DEPENDENCIES
  required_apps, shared DocTypes, hook contracts

STEP 4: DESIGN DATA MODEL
  DocTypes, relationships, naming conventions

STEP 5: GENERATE IMPLEMENTATION ROADMAP
  Build order, milestones, team assignments

See references/workflow.md for detailed steps.

Step 1: Requirement Analysis Matrix

Map each business requirement to Frappe mechanisms:

Requirement TypeFrappe MechanismExample
Data storageDocType"Track customer contracts"
Business rulesController/Server Script"Auto-calculate totals"
Approval flowWorkflow"Manager must approve orders >10k"
Scheduled tasksScheduler/hooks.py"Daily report email"
External syncIntegration/API"Sync with Shopify"
Custom UIClient Script/Page"Dashboard for warehouse"
ReportsScript Report/Query Report"Monthly sales by region"
PermissionsRole Permission"Sales team sees own data only"
Print outputPrint Format (Jinja)"Custom invoice layout"
Portal accessWebsite/Portal"Customer can view orders"

Step 2: App Boundary Decision Framework

Single App: Use When

  • Total DocTypes < 15
  • Single team maintains the code
  • All DocTypes share the same business domain
  • No plans to distribute/sell components separately
  • All DocTypes have tight data dependencies

Multiple Apps: Use When

  • Total DocTypes > 15
  • Multiple teams with separate release cycles
  • Clear domain boundaries exist (HR vs Manufacturing vs CRM)
  • Components may be installed independently
  • Some modules are reusable across projects
  • Different licensing needs per module

Decision Tree

HOW MANY DOCTYPES?
|
+-- < 15 total
|   +-- Single domain? --> SINGLE APP
|   +-- Multiple domains? --> Consider splitting
|
+-- 15-30 total
|   +-- Tight coupling between all? --> SINGLE APP (with modules)
|   +-- Clear domain boundaries? --> 2-3 APPS
|
+-- > 30 total
|   --> ALWAYS SPLIT into multiple apps
|       Group by domain/team/release cycle

See references/decision-tree.md for the complete decision framework.

Step 3: Cross-App Dependency Patterns

required_apps Declaration

ALWAYS declare dependencies explicitly in hooks.py:

python
# myapp/hooks.py
required_apps = ["frappe", "erpnext"]  # NEVER omit frappe

Dependency Rules

  • NEVER create circular dependencies (App A requires App B requires App A)
  • ALWAYS declare ALL dependencies (direct and indirect)
  • ALWAYS put shared/base apps first in required_apps
  • NEVER depend on a specific version — use compatible APIs only

Dependency Diagram Pattern

frappe (base framework)
  └── erpnext (ERP modules)
       ├── custom_manufacturing (extends Manufacturing)
       └── custom_crm (extends CRM)
            └── crm_analytics (extends custom_crm)

RULE: Dependencies flow DOWN only. Never up, never sideways.

Cross-App Communication Patterns

PatternMechanismUse When
Hook Eventsdoc_events in hooks.pyApp B reacts to App A's documents
Shared DocTypeLink fields to other app's DocTypesApps share reference data
API Callfrappe.call() to whitelisted methodLoose coupling between apps
Custom Fieldsfixtures with Custom FieldExtend another app's DocType without modifying it
Overrideextend_doctype_class (v16) or doc_eventsModify another app's behavior
Signalsfrappe.publish_realtime()Real-time notifications between apps

Step 4: Data Model Design

DocType Relationship Types

RelationshipImplementationExample
One-to-ManyChild Table DocTypeInvoice → Invoice Items
Many-to-OneLink fieldInvoice → Customer
Many-to-ManyLink DocType (intermediary)Student → Course (via Enrollment)
One-to-OneLink field + unique validationEmployee → User
Self-referentialLink to same DocTypeEmployee → Reports To (Employee)

Naming Conventions

ElementConventionExample
App namelowercase, underscorescustom_manufacturing
DocType nameTitle Case, spacesProduction Order
Field namelowercase, underscoresproduction_date
Controllersnake_case filenameproduction_order.py
ModuleTitle CaseManufacturing

Data Model Rules

  • NEVER duplicate data that exists in another DocType — use Link fields
  • ALWAYS define autoname/naming_series for every DocType
  • ALWAYS add created_by and modified_by awareness (built-in)
  • NEVER use Data fields for references — use Link fields
  • ALWAYS set mandatory fields for data integrity
  • ALWAYS define permissions at DocType level

App Composition Patterns

Pattern 1: Base + Vertical

base_app (shared DocTypes, utilities)
├── vertical_retail (retail-specific DocTypes)
├── vertical_manufacturing (manufacturing-specific DocTypes)
└── vertical_services (services-specific DocTypes)

Use when: Building industry-specific solutions on shared foundation.

Pattern 2: Core + Extensions

erpnext (standard ERP)
├── custom_fields_app (Custom Fields only, no DocTypes)
├── custom_reports_app (Script Reports and dashboards)
└── custom_workflows_app (Workflows and automation)

Use when: Extending ERPNext without modifying core. Keeps upgrades clean.

Pattern 3: Shared Utilities

frappe_utils (shared library: PDF generation, email templates, etc.)
├── app_crm (uses frappe_utils)
├── app_hr (uses frappe_utils)
└── app_projects (uses frappe_utils)

Use when: Multiple apps need the same utility functions.

Pattern 4: Marketplace App

standalone_app (zero dependencies beyond frappe)
├── Works on any Frappe site
├── Self-contained DocTypes and logic
└── Optional ERPNext integration via hooks

Use when: Building for distribution/sale on Frappe marketplace.

ERPNext Extension Patterns

Custom Fields vs Custom DocTypes vs Override

ApproachUse WhenProsCons
Custom FieldsAdding 1-10 fields to existing DocTypeSurvives upgrades, no codeLimited logic, UI clutter
Custom DocTypeNew business entity not in ERPNextFull control, clean designNo built-in ERPNext logic
Controller OverrideModifying existing ERPNext behaviorFull Python accessFragile on upgrades
Server ScriptSimple validation/automationNo custom app neededSandbox limitations
Client ScriptUI customizationNo custom app neededJS only, no server logic

Extension Decision Rules

  • ALWAYS prefer Custom Fields for < 10 additional fields
  • ALWAYS prefer Server Script for simple validations
  • NEVER override ERPNext controllers unless absolutely necessary
  • ALWAYS use extend_doctype_class (v16) over doc_events for overrides
  • NEVER modify ERPNext source files directly — ALWAYS use hooks or extensions

Common Architecture Mistakes

MistakeWhy It FailsCorrect Approach
Circular app dependenciesInstall/update breaksRestructure dependency tree
One mega-app with 50+ DocTypesUnmaintainable, slow testsSplit by domain into 3-5 apps
Duplicating ERPNext DocTypesData inconsistency, double maintenanceExtend with Custom Fields + hooks
No required_apps declarationSilent failures on fresh installALWAYS declare all dependencies
Shared database tables between appsTight coupling, migration conflictsUse Link fields and API calls
Modifying ERPNext source filesLost on every upgradeUse hooks, Custom Fields, extensions
No module organization within appFiles scattered, hard to navigateGroup DocTypes into modules
Hardcoded site/company namesBreaks on multi-site/multi-companyUse frappe.defaults and filters

Agent Output Format

ALWAYS produce architecture output in this format:

markdown
## Architecture Design

### Requirements Summary
| # | Requirement | DocTypes | Mechanism |
|---|------------|----------|-----------|

### App Structure
[Diagram showing apps and dependencies]

### App Inventory
| App | Module(s) | DocTypes | Dependencies |
|-----|-----------|----------|-------------|

### Data Model
| DocType | App | Key Fields | Relationships |
|---------|-----|------------|---------------|

### Cross-App Communication
| Source App | Target App | Mechanism | Trigger |
|-----------|-----------|-----------|---------|

### ERPNext Extensions
| Extension Type | Target DocType | Purpose |
|---------------|---------------|---------|

### Implementation Roadmap
| Phase | App(s) | Deliverables | Dependencies |
|-------|--------|-------------|-------------|

### Risk Assessment
| Risk | Mitigation |
|------|-----------|

### Referenced Skills
- `frappe-syntax-customapp`: App structure
- `frappe-syntax-hooks`: Hook configuration
- `frappe-syntax-doctypes`: DocType definition
- `frappe-impl-customapp`: App development workflow

See references/decision-tree.md for complete decision frameworks. See references/examples.md for architecture design examples.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Frappe Agent Architect AI skill do?

Use when designing multi-app Frappe architectures, deciding whether to split functionality into separate apps, or implementing cross-app communication patterns. Prevents monolithic app sprawl, circular dependencies between apps, and broken override chains. Covers multi-app architecture decisions, app dependency management, cross-app hooks, override patterns, when to split vs extend, shared DocType strategies. Keywords: architecture, multi-app, app splitting, cross-app, dependencies, override, extend, monolith, modular, how to structure frappe apps, when to split apps, app design, multi-app...

Why use Frappe Agent Architect on TypingMind?

Because you install it once and use it with any model. Frappe Agent Architect is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Frappe Agent Architect in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Impertio-Studio/Frappe_Claude_Skill_Package/tree/main/skills/source/agents/frappe-agent-architect. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Frappe Agent Architect?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Frappe Agent Architect?

As many as you like. As long as a model supports skills, you can use Frappe Agent Architect with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Frappe Agent Architect AI skill free?

Yes. It is published on GitHub by Impertio-Studio under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇