DevOps: Analyze Template

Analyze a Machina agent-template and provide a comprehensive overview of its components.

Trigger

• /mkn-devops:analyze-template
• "Analyze template [path]"
• "What's in this template?"
• "Overview of template"

Process

1. Identify Template Path

Ask user for the template location if not provided:

• Local: /Users/fernando/machina/dazn-templates/agent-templates/moderator-assistant
• Or template name to search in known repos

Known template repositories:

• /Users/fernando/machina/dazn-templates/agent-templates/
• /Users/fernando/machina/entain-templates/agent-templates/
• /Users/fernando/machina/machina-templates/agent-templates/

2. Discover All Files

List all YAML files in the template directory:

find <template-path> -type f -name "*.yml" | sort

Note special structures:

• agents/ - Agent definitions
• workflows/ - Workflow definitions
• prompts/ - Prompt definitions
• mappings/ - Data mappings
• scripts/ - Connector definitions (.yml + .py)
• configs/ - Setup/configuration workflows
• custom-*/ - Custom subdirectories (polls, quizzes, game, etc.)
• documents/ - Static document templates

3. Read Installation Manifest

Read _install.yml to extract:

Field	Location	Purpose
title	setup.title	Template display name
description	setup.description	What it does
version	setup.version	Semantic version
category	setup.category	Classification (special-templates, connectors, etc.)
estimatedTime	setup.estimatedTime	Installation time
features	setup.features	Feature bullet list
integrations	setup.integrations	Required connectors
status	setup.status	available, beta, deprecated
datasets	datasets[]	Installation order and types

4. Analyze Components

Agents (agents/*.yml)

For each agent, extract:

Property	YAML Path	Notes
Name	agent.name	Unique identifier
Title	agent.title	Display name
Description	agent.description	Purpose
Type	-	LLM-based or Orchestrator (has workflows:)
Model	agent.model or from connector	google-genai, machina-ai, etc.
System Prompt	agent.system-prompt	Reference to prompt name
Tools	agent.tools[]	Connectors used
Config	agent.context.config-*	frequency, temperature, etc.
Status	agent.context.status	active, inactive
context-agent	agent.context-agent	Input parameters with defaults
Workflows	agent.workflows[]	For orchestrator agents

Agent Types:

• LLM-based: Has model, system-prompt, tools
• Orchestrator: Has workflows[] list, coordinates multiple workflows

Workflows (workflows/*.yml, configs/*.yml, custom-*/*.yml)

For each workflow, extract:

Property	YAML Path	Notes
Name	workflow.name	Unique identifier
Title	workflow.title	Display name
Description	workflow.description	Purpose
Inputs	workflow.inputs	Input parameters
Outputs	workflow.outputs	Return values
Context Variables	workflow.context-variables	API keys, debugger, etc.
Tasks	workflow.tasks[]	Execution steps

Task Types:

Type	Purpose	Key Fields
document	CRUD on documents	action, filters, documents
connector	Call external service	connector.name, connector.command
prompt	LLM generation	connector, inputs, schema
mapping	Data transformation	inputs, outputs
function	Inline Python code	code
agent	Execute sub-agent	agent.name

Special Patterns:

• foreach - Loop over array: foreach.value, foreach.name, foreach.expr
• condition - Conditional execution: Python expression
• config.action - document actions: search, save, update, bulk-save, delete

Credentials Extraction:

Look for context-variables with values starting with $:

context-variables:
  google-genai:
    api_key: "$TEMP_CONTEXT_VARIABLE_GOOGLE_GENERATIVE_AI_API_KEY"
  machina-ai:
    api_key: "$TEMP_CONTEXT_VARIABLE_SDK_OPENAI_API_KEY"

Extract and deduplicate all $VARIABLE_NAME references across workflows to build the Required Credentials list.

Prompts (prompts/*.yml, custom-*/prompts.yml)

For each prompt, extract:

Property	YAML Path	Notes
Name	prompts[].name	Unique identifier
Title	prompts[].title	Display name
Description	prompts[].description	Purpose
Instruction	prompts[].instruction	System instruction text
Schema	prompts[].schema	JSON Schema for structured output
Languages	Schema properties	title_es, title_it, title_de, etc.

Language Support: Check for title_es, title_it, title_de, title_fr, title_pt, title_ja, title_he, title_ar in schema.

Mappings (mappings/*.yml)

For each mapping, extract:

• Name and Description
• Inputs - Source data
• Outputs - Transformed data
• Logic - Transformation expressions

Connectors (scripts/*.yml)

For each connector defined in the template, extract:

Property	YAML Path	Notes
Name	connector.name	Unique identifier
Title	connector.title	Display name
Description	connector.description	Purpose
Type	connector.filetype	pyscript, api, etc.
Filename	connector.filename	Python file (.py)
Status	connector.status	active, inactive
Commands	connector.commands[]	Available functions

Connector Usage Discovery

Scan all workflows for connector references:

grep -r "connector:" -A 3 <template-path> --include="*.yml" | grep -E "(name:|command:)"

Extract unique connectors and their commands:

• Connector name from connector.name
• Command from connector.command
• Workflow file where it's used

Connector-to-Secrets Mapping

Cross-reference discovered connectors with actual secret variable names used in the platform:

Connector	Secret Variable	Description
google-genai	$TEMP_CONTEXT_VARIABLE_GOOGLE_GENERATIVE_AI_API_KEY	Google Generative AI API Key
machina-ai	$TEMP_CONTEXT_VARIABLE_SDK_OPENAI_API_KEY	Machina AI / OpenAI API Key
google-storage	$MACHINA_CONTEXT_VARIABLE_GOOGLE_STORAGE_API_KEY	GCS Service Account JSON
google-storage	$MACHINA_CONTEXT_VARIABLE_GOOGLE_STORAGE_BUCKET_NAME	GCS Bucket Name
sportradar-soccer	$TEMP_CONTEXT_VARIABLE_SPORTRADAR_SOCCER_V4_API_KEY	SportRadar Soccer API Key
sportradar-nba	$TEMP_CONTEXT_VARIABLE_SPORTRADAR_NBA_API_KEY	SportRadar NBA API Key
sportradar-nfl	$TEMP_CONTEXT_VARIABLE_SPORTRADAR_NFL_API_KEY	SportRadar NFL API Key
openai	$TEMP_CONTEXT_VARIABLE_OPENAI_API_KEY	OpenAI API Key
groq	$TEMP_CONTEXT_VARIABLE_GROQ_API_KEY	Groq API Key
wordpress	$TEMP_CONTEXT_VARIABLE_WORDPRESS_BEARER_TOKEN	WordPress Bearer Token
api-football	$TEMP_CONTEXT_VARIABLE_API_FOOTBALL_API_KEY	API-Football Key
bwin	$TEMP_CONTEXT_VARIABLE_BWIN_ACCESS_ID	BWin Access ID
bwin	$TEMP_CONTEXT_VARIABLE_BWIN_ACCESS_ID_TOKEN	BWin Access Token
vertex-ai	$TEMP_CONTEXT_VARIABLE_VERTEX_AI_CREDENTIAL	Vertex AI Service Account
vertex-ai	$TEMP_CONTEXT_VARIABLE_VERTEX_AI_PROJECT_ID	Vertex AI Project ID
temp-downloader	None	Public URL access only
storage	None	Local storage

Note: Secret names follow the pattern $TEMP_CONTEXT_VARIABLE_<SERVICE>_<KEY_TYPE>

Reference: See Connectors Catalog for complete list

5. Generate Report

Use this template structure:

# Template Analysis: {template-name}

## Overview
| Field | Value |
|-------|-------|
| **Title** | {from _install.yml} |
| **Description** | {description} |
| **Version** | {version} |
| **Category** | {category} |
| **Estimated Time** | {estimatedTime} |

## Features
- Feature 1
- Feature 2
- ...

## Dependencies

### Required Connectors
| Connector | Purpose |
|-----------|---------|
| connector-1 | Description |
| connector-2 | Description |

### External Dependencies
- GCS bucket: {url}
- External API: {name}

### Required Secrets
| Secret | Purpose |
|--------|---------|
| SECRET_NAME | What it's used for |

### Required Credentials (from context-variables)
| Variable | Connector | Workflow |
|----------|-----------|----------|
| `$TEMP_CONTEXT_VARIABLE_GOOGLE_GENERATIVE_AI_API_KEY` | google-genai | workflow-name |
| `$TEMP_CONTEXT_VARIABLE_SDK_OPENAI_API_KEY` | machina-ai | workflow-name |

### Connectors Used
| Connector | Commands | Workflows |
|-----------|----------|-----------|
| google-genai | invoke_prompt | custom-* |
| machina-ai | invoke_prompt | custom-* |
| temp-downloader | invoke_download, invoke_read_json | sync-*, ros-engine |
| google-storage | invoke_upload | generate-and-upload |

### Complete Secrets Checklist
| Secret Variable | Connector | Source | Required |
|-----------------|-----------|--------|----------|
| `$TEMP_CONTEXT_VARIABLE_GOOGLE_GENERATIVE_AI_API_KEY` | google-genai | context-variables | Yes |
| `$TEMP_CONTEXT_VARIABLE_SDK_OPENAI_API_KEY` | machina-ai | context-variables | Yes |
| `$MACHINA_CONTEXT_VARIABLE_GOOGLE_STORAGE_API_KEY` | google-storage | connector | Conditional |
| `$MACHINA_CONTEXT_VARIABLE_GOOGLE_STORAGE_BUCKET_NAME` | google-storage | connector | Conditional |

## Components Summary

| Type | Count | Files |
|------|-------|-------|
| Agents | X | file1.yml, file2.yml |
| Workflows | X | file1.yml, ... |
| Prompts | X | file1.yml |
| Mappings | X | file1.yml |
| Connectors | X | file1.yml |

---

## Agents Detail

### {agent-name}
| Property | Value |
|----------|-------|
| **Description** | ... |
| **Type** | LLM-based / Orchestrator |
| **Model** | google-genai (gemini-2.0-flash) |
| **Status** | active / inactive |

**Parameters (context-agent):**
- `param1` - Description (default: value)
- `param2` - Description

**Orchestrated Workflows:** (if orchestrator)
1. workflow-1 - Description
2. workflow-2 - Description (condition: ...)

---

## Workflows Detail

### Core Workflows

#### {workflow-name}
| Property | Value |
|----------|-------|
| **Description** | ... |
| **Inputs** | param1, param2 |
| **Outputs** | output1, output2 |
| **Tasks** | X steps |

**Task Flow:**

task-1 (document) → task-2 (connector) → task-3 (document)

### Setup Workflows
- workflow-setup-dev - Description
- workflow-setup-prd - Description

### Sync Workflows
- workflow-sync-x - Description

---

## Connector Detail

### {connector-name}
| Property | Value |
|----------|-------|
| **Type** | pyscript |
| **File** | filename.py |
| **Status** | active |

**Commands:**
| Command | Function |
|---------|----------|
| Command Name | function_name |

---

## Prompts Detail

### {prompt-name}
| Property | Value |
|----------|-------|
| **Purpose** | What it generates |
| **Used by** | workflow or agent name |
| **Languages** | EN, ES, IT, DE, FR, PT, JA, HE, AR |
| **Schema** | SchemaName (key fields) |

---

## Data Flow Diagram

                ┌─────────────────────────────────────────┐
                │           External Source                │
                └─────────────┬───────────────────────────┘
                              │
        ┌─────────────────────┼─────────────────────┐
        ▼                     ▼                     ▼
┌───────────────┐    ┌───────────────┐    ┌───────────────┐
│   workflow-1  │    │   workflow-2  │    │   workflow-3  │
└───────┬───────┘    └───────┬───────┘    └───────┬───────┘
        │                    │                    │
        ▼                    ▼                    ▼
┌───────────────┐    ┌───────────────┐    ┌───────────────┐
│   document-1  │    │   document-2  │    │   document-3  │
└───────────────┘    └───────────────┘    └───────────────┘

---

## Installation Order

From `_install.yml` datasets:
1. type: connector - path
2. type: prompts - path
3. type: workflow - path
4. type: agent - path

6. Output Options

Ask user preference:

• Console: Display report in terminal (default)
• File: Save to {template-name}-analysis.md

Example Analysis

See actual analysis of dazn-runofshow template for reference format.

Tips

• Use this skill before installing a template to understand its requirements
• Check required connectors are available in target environment
• Verify secrets are configured before installation
• Review agent models to ensure API keys are available
• Look for foreach patterns - they indicate batch processing
• Check condition fields - they show branching logic
• Credentials: Scan all context-variables for $TEMP_CONTEXT_VARIABLE_* patterns - these must be configured in the environment secrets
• Common credential patterns:
  • $TEMP_CONTEXT_VARIABLE_GOOGLE_GENERATIVE_AI_API_KEY → Google AI
  • $TEMP_CONTEXT_VARIABLE_SDK_OPENAI_API_KEY → Machina AI / OpenAI
  • $TEMP_CONTEXT_VARIABLE_SPORTRADAR_API_KEY → SportRadar