#Open Source Tool1 utente parla di questo argomento

 Introducing MetaDelta

: A Salesforce CLI Plugin Built for DevOps Teams 

As Salesforce DevOps engineers, we've all faced the same challenges:

 ❌ "Who modified this metadata?" 

 ❌ "What exactly was deployed?" 

 ❌ "Can I generate a package.xml only for what changed?" 

 ❌ "How do I audit Salesforce Core and Vlocity/DataPacks together?" 

 ❌ "How can I automate manual Salesforce Setup tasks?"

That's why I built MetaDelta, an open-source Salesforce CLI plugin focused on metadata auditing, deployment analysis, manifest generation, and manual task automation. 

🔍 Current Features

📊 Metadata Auditing

Identify metadata modified by a specific user within a configurable time window.

•  Salesforce Core support 
•  Vlocity / Industries support 
•  User-based filtering 
•  Change tracking across environments 
•  Deployment investigation 

Perfect for answering:

📦 Smart Manifest Generation

Automatically generate deployment manifests from detected changes.

Supports:

•  package.xml 
•  package.yaml 
•  Salesforce Core Metadata 
•  Vlocity DataPacks 

Useful for:

•  Hotfix deployments 
•  UAT packaging 
•  Release preparation 
•  Partial deployments 

🔎 Deployment & Change Analysis

Analyze deployment activity and understand:

•  What was deployed 
•  When it was deployed 
•  Which components were affected 
•  Who performed the deployment 

A feature designed to reduce investigation time during incidents and release validation.

🤖 Manual Salesforce Setup Automation

One of the newest capabilities in MetaDelta.

Record and replay manual Salesforce Setup activities using Playwright-based automation.

Examples:

✅ Enable Salesforce features 

 ✅ Configure Agentforce settings 

 ✅ Navigate complex Setup pages 

 ✅ Standardize manual deployment steps 

 ✅ Reproduce setup tasks across multiple orgs

This helps transform traditionally undocumented manual activities into repeatable automation.

🎥 Record Once, Replay Everywhere

MetaDelta can:

1.  Record manual Salesforce actions 
2.  Generate executable automation scripts 
3.  Replay them later across environments 

Great for:

•  Sandbox setup 
•  Release preparation 
•  Environment standardization 
•  Compliance processes

🎯 Built For

•  Salesforce DevOps Engineers 
•  Release Managers 
•  Salesforce Architects 
•  Vlocity / Salesforce Industries Teams 
•  CI/CD Engineers 
•  Enterprise Salesforce Programs

⭐ Why MetaDelta?

Most Salesforce tools focus on deployment.

MetaDelta focuses on:

✅ Understanding change 

 ✅ Auditing metadata 

 ✅ Generating deployment artifacts 

 ✅ Automating manual setup activities 

 ✅ Supporting both Salesforce Core and Vlocity ecosystems 

🔗 GitHub Repository

If you're interested in Salesforce DevOps, metadata governance, release management, or Salesforce Industries automation, I'd love your feedback.

MetaDelta GitHub Repository

⭐ Star the project if you find it useful.

Contributions, ideas, feature requests, and feedback are always welcome. 

I open-sourced 3 tools I kept rebuilding across Salesforce projects. Maybe they save you some time too :)

After 15 years of Salesforce implementations, I noticed I was solving the same three problems on almost every project. Data migrations that break in the same ways. Trigger code that turns into spaghetti. AI agents that need 500 lines of glue code just to read a Contact record. So I packaged up the patterns that actually survived production and put them on GitHub. All free, all MIT/Apache licensed. Take what's useful, ignore what isn't. Here's what's in there:

1. Salesforce AI Agent Framework (SAAF) https://github.com/sandiphpatel87/salesforce-ai-agent-framework  
   If you've tried building AI agents that work with Salesforce, you know the pain. LangChain and CrewAI don't understand governor limits. They don't know about FLS. They can't invoke a Flow or navigate a lookup chain. SAAF is a Python framework with agents that actually understand Salesforce. It ships with three pre-built agents (Lead Qualification, Case Routing, Opportunity Scoring) and works with Claude, GPT, or local models through Ollama. The part I'm most excited about: it includes an MCP Server. If you're using Claude Desktop or Cursor, you can connect it directly to your Salesforce org. Ask Claude "show me all opportunities closing this month over $50K" and it writes the SOQL, runs it, and gives you the results. No code needed on your end. Every agent action gets audit logged, there's a dry-run mode so you can preview changes before committing, and it tracks governor limits automatically so you don't blow through your API calls. Stack: Python 3.10+, simple-salesforce, MCP SDK. MIT license.

2. CRM Data Migration Toolkit https://github.com/sandiphpatel87/crm-data-migration-toolkit  
   I've done 100+ data migrations at this point. The same things break every time: phone numbers in 15 different formats, state names that aren't abbreviations, orphaned Contacts with no Account, picklist values that don't match between systems. This toolkit handles the unglamorous work. It has 13 built-in transforms (phone normalization, state abbreviation, date parsing, HTML stripping, etc.), a validation framework that catches problems before you touch Salesforce, and deduplication logic. It also ships with pre-built templates. Right now there's a HubSpot-to-Salesforce template that maps Companies to Account, Contacts to Contact, and Deals to Opportunity with the correct stage mapping. If you're doing a HubSpot migration next quarter, this saves you a few days. The whole thing runs in dry-run mode by default so you see exactly what would happen before a single record gets created. Stack: Python 3.10+, simple-salesforce. Apache 2.0 license.

3. Salesforce Process Automation Recipes https://github.com/sandiphpatel87/salesforce-process-automation-recipes  
   This one is less of a tool and more of a cookbook. It's a collection of automation patterns I've used across different orgs, each with the actual code, a writeup of the business scenario, and the gotchas that only show up in production. What's in there so far:

Trigger Handler Framework with recursion guards and per-user bypass (the "one trigger per object" pattern done right) Queueable Chain for processing 100K+ records without hitting limits REST Callout with Circuit Breaker so your integration doesn't take down the org when the external API goes sideways Slack Notification Engine with Block Kit formatting AI Case Classifier that auto-categorizes incoming cases using an LLM Lead Round Robin Flow with timezone awareness and vacation handling Opportunity Stage Validation with configurable gates via Custom Metadata Case Escalation Timer for SLA breach automation

Each recipe includes the business problem it solves, not just the code. I tried to write the docs I wish existed when I was figuring this stuff out. Stack: Apex, Flows, LWC. MIT license.

Contributing All three repos have CONTRIBUTING.md files and issue templates. If you've solved similar problems and want to add your patterns, PRs are welcome. I'm especially looking for:

Migration templates for other CRMs (Dynamics, Zoho, Pipedrive, Freshsales) Industry-specific agent recipes (Healthcare, Financial Services) Performance benchmarks from high-volume orgs

If any of this is useful to you, that's great. If you have feedback on what could be better, I'd rather hear it. Drop a comment or open an issue.  

Your Friend, Sandip 

Agentforce NOW AMA: Build with React and Salesforce Multi-Framework | May 27, 2026

Join the Salesforce Multi-Framework product team for a live AMA where we explore how to build production-ready apps with React. You'll see how to build dynamic apps that are connected to Salesforce data, secured by your org's existing permissions, and powered by the Agentforce 360 Platform.  

Our experts will walk you through the roadmap and answer your questions on: 

• Developing with React: Build customer portals, employee-facing apps, and micro-frontends using React's expansive open-source ecosystem
• Integrating with Agentforce Vibes: Go from natural-language prompt to working React app with Agentforce Vibes 2.0
• Scaling on the Headless 360 Platform: Query records and invoke Apex from React with automatic authentication and access to the business logic, workflows, and compliance rules built into your org

Whether you're already building with React or just getting started, this is the session for you. 

Register and join us live: https://www.linkedin.com/events/7456009299094339585/

Can't make it live? All AMAs are recorded and available on demand on the Salesforce Developers YouTube channel. 

#React Native #Open Source Tool #Prompts #Headless #Apex   

Free Production-Ready AI Web + Voice Chat → Salesforce Lead/Case in <60 seconds (AWS + Lex + Connect + Bedrock + Step Functions) 

I just open-sourced the entire Atlas Engine — a $150k+ equivalent sales acceleration platform — 100% free on GitHub.

Live demo you can try right now: https://atlasengine.net

GitHub (full code + one-click deploy):

What it does in real time:

• User types in web chat → instantly creates Salesforce Lead
• AI (Bedrock using Claude 3.5 Sonnet) generates a personalized script
• Amazon Connect calls their phone in <60 seconds
• They talk naturally to the AI voice bot
• Can ask “delete my info” → instantly deletes the Salesforce record (GDPR ready)
• Can say “talk to the architect” → creates a Case routed to me with full transcript

Everything is serverless, production-grade, and deploys in 15–30 minutes with ./deploy.sh

(Full CloudFormation one-click also included)

Includes:

• 8 Lambdas + Step Functions orchestration
• Full Salesforce JWT integration
• Bedrock dynamic scenario generation
• Transcribe + summarization back to Lead
• Delete-my-info compliance built in

Try the live demo now at atlasengine.net

— give it your real number and watch it call you instantly. 

Feedback, stars, and forks very welcome! I hope you build something cool with The Atlas Engine!