Kindling

The AI-Native Development Framework

General AI models are brilliant, but they don’t live in the unique world of Salesforce. They struggle with governor limits, outdated LWC syntax, and the complexities of the Metadata API.

Kindling is a developer-centric toolkit built to bridge the gap between Large Language Models and the Salesforce platform. It is a specialized suite of .cursorrules and prompt templates optimized for the Cursor code editor, designed to make your AI-pair programmer an expert Salesforce Architect.

---

Why Kindling?

⚡️ From "Prompting" to "Building"

Standard AI often hallucinates Salesforce methods or uses inefficient patterns. Kindling provides a structured framework that "primes" the AI with your specific project standards, ensuring that every line of code generated follows modern best practices and your organization's specific style guide.

🏛️ Architectural Integrity

Framework-driven development, not just code snippets.

Kindling transforms Cursor from a simple editor into a high-fidelity architectural partner. Our rulesets are pre-configured to enforce platform best practices across your entire project, supporting Agentic Workflows that allow the AI to autonomously plan and execute multi-file refactors. Whether you are using a standard Trigger Framework, optimizing SOQL for large data volumes, or enforcing modern LWC wire service patterns, Kindling provides the guardrails.

Because the framework is optimized for frontier reasoning models, it leverages advanced "Chain of Thought" logic to validate your organization's specific design patterns before a single file is updated, ensuring your technical debt stays low and your deployment success rate stays high.

• Trigger Frameworks: Automatically align logic with your existing trigger handlers.
• LWC Best Practices: Enforce modern reactive properties and efficient wire service usage.
• SOQL Optimization: Prevent "Non-selective query" errors before they are even written.

🎯 Context-Aware Engineering

Treat your metadata as a single, searchable brain.

Generic AI models struggle because they see your code in isolation. By leveraging Cursor’s Agent Mode (Composer) alongside Kindling’s optimized rulesets, the AI gains a holistic understanding of your entire force-app directory.

Kindling maps the deep dependencies between your Custom Objects, JS-Meta XMLs, and Apex Controllers, ensuring the AI understands how a change in one file impacts the rest of your architecture. This metadata-first approach is Optimized for Frontier Reasoning Models, allowing them to navigate complex Salesforce logic with surgical precision—drastically reducing hallucinations and ensuring every line of code is grounded in your actual org schema.

---

Optimized for Frontier Reasoning Models

Kindling’s architecture is specifically tuned to leverage the advanced logic of frontier reasoning models. By providing structured "chain-of-thought" instructions, it ensures that high-reasoning LLMs correctly navigate complex Apex logic and multi-object metadata relationships without losing architectural integrity.

---