Call use_skill as your FIRST and ONLY action when the user asks to document, add descriptions to, annotate, or add doc:description attributes to Mule XML files, flows, components, or connectors. Do not read project files first — this skill provides instructions for when to read files. Covers documenting all Mule elements in src/main/mule including flows, sub-flows, configs, listeners, processors, transforms, error handlers, and connectors. When you call use_skill, it must be the only tool call in that response.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Um comando direto ignora o prompt de revisão. Verifique a origem antes de executá-lo.
Instruções da origem · Visualização somente leitura
name
generate-doc-description
description
Call use_skill as your FIRST and ONLY action when the user asks to document, add descriptions to, annotate, or add doc:description attributes to Mule XML files, flows, components, or connectors. Do not read project files first — this skill provides instructions for when to read files. Covers documenting all Mule elements in src/main/mule including flows, sub-flows, configs, listeners, processors, transforms, error handlers, and connectors. When you call use_skill, it must be the only tool call in that response.
metadata
{"author":"mule-dx-tooling","version":"1.0.0"}
Overview
You are an XML documentation expert for MuleSoft. Add doc:description attributes to XML elements in Mule configuration files.
Scan the Mule project for configuration XML files in src/main/mule/ and add or update doc:description attributes on elements that are missing them or have inaccurate descriptions.
Rules
doc:description: A meaningful 1-2 sentence purpose description. MUST NOT exceed 150 characters.
Preserve ALL existing XML structure, attributes, CDATA blocks, and content exactly as-is
Do NOT modify, add, or remove any non-doc attributes or elements
CRITICAL: NEVER call Read and Write in the same turn - When you call Read tool, do NOT call Write tool in that same response. Read in one turn, analyze and prepare changes, then Write in the next turn. This is a hard requirement - Read and Write are always separate turns.
Important Rules
Complete each file fully in ONE Write operation - do NOT make edits in batches
Do NOT modify doc:name attributes - only work with doc:description
Do NOT modify, add, or remove any non-doc attributes or elements
Preserve ALL existing XML structure, attributes, CDATA blocks, and content exactly as-is
Add doc:description to ALL elements listed in section B - including loggers, processors, connectors, configs, etc.
Do NOT change existing accurate descriptions - only update if they're vague or incorrect
Maintain XML formatting - preserve indentation and structure
Be specific - avoid generic phrases like "processes data" or "handles request"
Enforce 150 character limit - all descriptions must be 150 characters or less
Error Handling
If a file cannot be read, skip it and report the error
If XML parsing seems problematic, inform the user
If uncertain about a component's purpose, generate a conservative description or ask the user
Step 1: Verify Project Structure
Check that src/main/mule directory exists in the current working directory
If not found, inform the user this doesn't appear to be a Mule application project
Step 2: Find All Mule XML Files
Use Glob to find all XML files: src/main/mule/**/*.xml
If no files are found, inform the user and exit
Step 3: Process Each File Completely
For each XML file found:
A. Read the entire file FIRST
CRITICAL: You MUST read the actual file content before making any changes.CRITICAL: When you call Read tool, do NOT call Write tool in the same turn. Read only.
Use Read tool to load the complete file contents
Do NOT make assumptions about what's in the file
Do NOT use content from other files as a template
Examine the actual namespace declarations, element names, attribute order, and structure
Verify the file structure matches expectations before proceeding
After reading, proceed to step B in this same turn (analysis), but save Write for a later turn
B. Analyze and explicitly list all flows/components
IMPORTANT: Before making ANY changes, explicitly list out what you found to prove you read the file:
Search through the file and count all major components:
Count all <flow> elements and list each by name attribute
Count all <sub-flow> elements and list each by name attribute
Count all <batch:job> elements and list each by jobName attribute
Count global configs (<*:config>, <*:listener-config>, <configuration-properties>)
Count global error handlers (<error-handler> at top level, outside flows)
Study the template above to understand the level of detail and style expected for each element type. Note how descriptions are:
Configuration properties and connector configs: mention what system and auth method
Connection elements: specify auth type and credential types
Flows and listeners: describe business process and what triggers them
Scheduling strategies: state polling frequency in clear terms
Loggers: explain what specific information is logged and why
Choice routers and when/otherwise: describe conditions and routing logic
Transforms: indicate source data, target format, and transformation purpose
Connector operations: state what operation and which system
Error handlers: explain what errors are caught and what action is taken
D. Make ALL edits to the file in ONE Write operation
CRITICAL: Use ONE Write tool call per file, regardless of file size.
Work with the ACTUAL file content you read in step A
Add doc:description attributes to ALL applicable elements in the file
Preserve ALL existing content exactly:
Namespace declarations in their original order
Element names and attribute names exactly as they appear
All existing attributes (doc:id, doc:name, etc.)
CDATA blocks, comments, whitespace, and indentation
Line breaks and formatting
Single Write Strategy:
Build the complete modified file content with ALL doc:description additions
Use ONE Write tool call to save the complete modified file back to disk
This ensures ONE approval prompt per file, regardless of file size
No batching, no multiple writes - just one complete operation
Why this single Write approach:
Simplest strategy - one write per file
ONE approval prompt per file regardless of size
Complete atomic operation - either all changes apply or none
Easier to review all changes at once
No risk of partial updates
Step 4: Present Changes for Approval
After completing ALL edits for a file:
Summarize what doc:description attributes were added (count by element type)
Highlight any elements that were skipped and why
The Write tool will prompt for user approval automatically
If approved, move to the next file
If denied, ask what needs adjustment and re-read the file to verify current state
Step 5: Repeat for All Files
Continue this process for each XML file in the src/main/mule/ directory.
Step 6: Final Summary
After all files are processed, provide:
Number of files scanned
Number of files modified
Total number of doc:description attributes added
Total number of doc:description attributes updated
"${salesforce.username}"
password
"${salesforce.password}"
doc:description
"Basic authentication with username and password credentials"
</salesforce:sfdc-config>
<!-- NetSuite Configuration -->
<netsuite:configname="netsuiteConfig"doc:name="NetSuite Config"doc:description="Configures connection to NetSuite using token-based authentication">
<netsuite:token-based-authentication-connectionconsumerKey="${netsuite.consumerKey}"consumerSecret="${netsuite.consumerSecret}"tokenId="${netsuite.tokenId}"tokenSecret="${netsuite.tokenSecret}"account="${netsuite.account}"doc:description="Token-based authentication with consumer and token credentials" />
</netsuite:config>
<!-- Main Integration Flow -->
<flowname="salesforce-to-netsuite-order-flow"doc:name="Salesforce to NetSuite Order Flow"doc:description="Syncs Closed Won Salesforce opportunities to NetSuite as sales orders">
<!-- Salesforce Opportunity Listener -->
<salesforce:modified-object-listenerobjectType="Opportunity"config-ref="salesforceConfig"doc:name="Listen for Modified Opportunities"doc:description="Polls Salesforce every 60 seconds for modified Opportunity records">
<scheduling-strategydoc:description="Defines polling frequency for the listener">
<fixed-frequencyfrequency="60000"doc:description="Polls every 60,000 milliseconds (1 minute)" />
</scheduling-strategy>
</salesforce:modified-object-listener>
<loggerlevel="INFO"message="Opportunity modified: #[payload.Id] - Stage: #[payload.StageName]"doc:name="Log Opportunity Update"doc:description="Logs the ID and stage of the modified Opportunity record" />
<!-- Filter for Closed Won Opportunities -->
<choicedoc:name="Check if Closed Won"doc:description="Routes the flow based on whether the Opportunity is Closed Won">
<whenexpression="#[payload.StageName == 'Closed Won']"doc:description="Processes opportunities with StageName equal to 'Closed Won'">
<loggerlevel="INFO"message="Processing Closed Won Opportunity: #[payload.Name] (Amount: #[payload.Amount])"doc:name="Log Closed Won"doc:description="Logs the name and amount of the Closed Won Opportunity being processed" />
<!-- Transform Salesforce Opportunity to NetSuite Order -->
<ee:transformdoc:name="Map Opportunity to NetSuite Order"doc:description="Transforms Salesforce Opportunity data into NetSuite sales order format using DataWeave">
<ee:messagedoc:description="Defines the message transformation for the payload">
<ee:set-payloaddoc:description="Maps Opportunity fields to NetSuite order structure with customer and line items">
</ee:set-payload>
</ee:message>
</ee:transform>
<loggerlevel="INFO"message="Transformed payload for NetSuite: #[output application/json --- payload]"doc:name="Log Transformed Payload"doc:description="Logs the transformed payload in JSON format for debugging" />
<!-- Create Order in NetSuite -->
<netsuite:addconfig-ref="netsuiteConfig"doc:name="Create NetSuite Order"doc:description="Creates a new sales order record in NetSuite using the add operation" />
<loggerlevel="INFO"message="NetSuite Order created successfully: #[payload]"doc:name="Log Success"doc:description="Logs successful creation of NetSuite order with response payload" />
</when>
<otherwisedoc:description="Handles opportunities that are not in Closed Won stage">
<loggerlevel="INFO"message="Opportunity not in Closed Won stage, skipping. Current stage: #[payload.StageName]"doc:name="Log Skipped"doc:description="Logs that the Opportunity is being skipped with its current stage" />
</otherwise>
</choice>
<!-- Error Handling -->
<error-handlerdoc:description="Handles errors that occur during the flow execution">
<on-error-propagatetype="ANY"doc:description="Catches all error types and propagates them after logging">
