From 539cef2510f4f90382108535ab2198590ad3493d Mon Sep 17 00:00:00 2001 From: Oliver Eggert Date: Thu, 9 Apr 2026 10:25:00 -0700 Subject: [PATCH 1/4] initial draft --- resources/dev-tools/ai-tools.md | 52 +++++++++++++++++++++++++++++++++ sidebars.yaml | 1 + 2 files changed, 53 insertions(+) create mode 100644 resources/dev-tools/ai-tools.md diff --git a/resources/dev-tools/ai-tools.md b/resources/dev-tools/ai-tools.md new file mode 100644 index 0000000000..930477743f --- /dev/null +++ b/resources/dev-tools/ai-tools.md @@ -0,0 +1,52 @@ +--- +seo: + description: Accelerate development on the XRPL with AI tools. +--- +# AI Tools + +The industry has shifted sharply toward _Agentic Workflows_ to accelerate development, automate integrations, and deploy secure solutions on the XRP Ledger. To meet this growing segment of developers, we provide AI tools to reduce hallucinations, automate tasks, and improve code generation. + + +## Model Context Protocol (MCP) Servers + +Ensure your LLM uses the latest infoThe direct data pipe for AI. The Model Context Protocol (MCP) allows your AI tools to skip the "search and scrape" phase. By connecting your IDE to the XRPL MCP server, you provide your LLM with a real-time, version-accurate stream of markdown files from the official repository. +Connect to the Context7 agentic registry to import the full XRPL knowledge base. +CTA: [Context7 MCP] + + +## Agent Intelligence & Skills (SKILLS.md) +Logic-level guardrails for autonomous agents. Raw data is not enough for secure financial deployments. We are introducing official SKILLS.md files—standardized behavioral instructions that act as a "firmware update" for your AI models. +Deterministic Logic: Forces LLMs to follow strict transaction fee handling and sequence logic. +Compliance & Security: Teaches models to prioritize multisig configurations and check for active network amendments (like XLS-30 or XLS-66) before proposing code. +Status: [Coming Soon] + + +## Site Optimizations + +### llms.txt + +Any AI that searches the website has access to an `llms.txt` file at the site directory root. This file serves as a curated index of content, ensuring it can find relevant information quickly. + +### Context-Optimization + +Markdown files are the best format to feed an LLM context. Every page on the doc site hosts an `.md` version, which is accessible from the `Copy` dropdown at the top of the page. The full list of copy commands include: + +- Copying the contents of the raw `.md` file. +- A link to the `.md` version of the page. +- A button that adds the page URL to a chat window in either `Claude` or `ChatGPT` chat windows. +- A one-click MCP server setup in either `VS Code` or `Cursor`. + + + Every page has a markdown version + +## RAG-Optimized Infrastructure +The site architecture is being optimized for Retrieval-Augmented Generation (RAG). +Semantic Precision +The core pages are undergoing a semantic audit to eliminate ambiguous language and implement self-contained chunking. +When your RAG pipeline retrieves a snippet, it contains all the context necessary for the LLM to understand the logic without needing to crawl adjacent pages. +The llms.txt Standard +The site hosts a specialized llms.txt index. This serves as a high-density "map" for crawlers, ensuring your AI assistants are always referencing the official source of truth. +Overall Developer Experience +We have removed the friction between the documentation and the codebase with features designed for high-stakes environments. +Copy for LLM: Powered by Redocly, our code samples feature a specialized "Copy for LLM" button. This removes UI clutter and formats the code with the metadata headers LLMs need for perfect prompt injection. +Hybrid Tutorials: Our guides are built as "Modular Lego Blocks." They provide the narrative "Why" for human reviewers and the deterministic "How" for AI agents to execute without confusion. diff --git a/sidebars.yaml b/sidebars.yaml index 912e1aa250..829a0ffc96 100644 --- a/sidebars.yaml +++ b/sidebars.yaml @@ -750,6 +750,7 @@ page: resources/dev-tools/index.page.tsx expanded: false items: + - page: resources/dev-tools/ai-tools.md - label: RPC Tool labelTranslationKey: sidebar.resources.dev-tools.rpc-tool page: resources/dev-tools/rpc-tool.page.tsx From 3d3ac6adb31a178a737c3d3ed59cf3b4320563cb Mon Sep 17 00:00:00 2001 From: Oliver Eggert Date: Thu, 9 Apr 2026 16:52:21 -0700 Subject: [PATCH 2/4] second iteration --- resources/dev-tools/ai-tools.md | 59 ++++++++++++++++----------------- 1 file changed, 28 insertions(+), 31 deletions(-) diff --git a/resources/dev-tools/ai-tools.md b/resources/dev-tools/ai-tools.md index 930477743f..7e9941b843 100644 --- a/resources/dev-tools/ai-tools.md +++ b/resources/dev-tools/ai-tools.md @@ -4,49 +4,46 @@ seo: --- # AI Tools -The industry has shifted sharply toward _Agentic Workflows_ to accelerate development, automate integrations, and deploy secure solutions on the XRP Ledger. To meet this growing segment of developers, we provide AI tools to reduce hallucinations, automate tasks, and improve code generation. +AI tools help you build on the XRP Ledger faster by giving AI models access to up-to-date XRPL documentation, SDK references, and code samples. ## Model Context Protocol (MCP) Servers -Ensure your LLM uses the latest infoThe direct data pipe for AI. The Model Context Protocol (MCP) allows your AI tools to skip the "search and scrape" phase. By connecting your IDE to the XRPL MCP server, you provide your LLM with a real-time, version-accurate stream of markdown files from the official repository. -Connect to the Context7 agentic registry to import the full XRPL knowledge base. -CTA: [Context7 MCP] +AI models are limited by their training data, which may not include the latest XRPL features or SDK updates. MCP servers solve this by giving your AI real-time access to current documentation through a standardized interface. Instead of relying on potentially outdated training data, your AI can query an MCP server for accurate, up-to-date context. Below is a list of available MCP servers: + +- **Context7**: A large collection of searchable repos and websites. XRPL-relevant info includes: + - [xrpl.org](https://xrpl.org/docs) + - [opensource.ripple.com](https://opensource.ripple.com/) + - [docs.xrplevm.org](https://docs.xrplevm.org/) + - [XRPL JavaScript SDK](https://github.com/xrplf/xrpl.js) + - [XRPL Python SDK](https://github.com/xrplf/xrpl-py) + - [XRPL Go SDK](https://github.com/xrplf/xrpl-go) +- **xrpl.org MCP Server**: Search across documentation on xrpl.org for a given query. -## Agent Intelligence & Skills (SKILLS.md) -Logic-level guardrails for autonomous agents. Raw data is not enough for secure financial deployments. We are introducing official SKILLS.md files—standardized behavioral instructions that act as a "firmware update" for your AI models. -Deterministic Logic: Forces LLMs to follow strict transaction fee handling and sequence logic. -Compliance & Security: Teaches models to prioritize multisig configurations and check for active network amendments (like XLS-30 or XLS-66) before proposing code. -Status: [Coming Soon] +## SKILLS.md + +A SKILLS.md file provides behavioral instructions for AI models working with XRPL code. It defines domain-specific rules — like transaction validation, fee handling, and security best practices — so the AI follows correct patterns without you having to explain them each time. + +- **[XRPL Development Skill for Claude Code](https://github.com/XRPL-Commons/xrpl-dev-skills)**: A comprehensive Claude Code skill for modern XRP Ledger development, provided by XRPL Commons. This skill uses Claude Code's progressive disclosure pattern. The main `SKILL.md` provides core guidance, and Claude reads specialized markdown files only when needed for specific tasks. +- **[Generate Release Notes](https://github.com/XRPLF/xrpl-dev-portal/blob/master/.claude/skills/generate-release-notes/SKILL.md)**: This skill generates a draft release notes blog post for new versions of `rippled`. This is intended for those contributing to the XRPL documentation site. ## Site Optimizations +### AI Search + +We include an AI chatbot on xrpl.org, opensource.ripple.com, and docs.xrplevm.org to provide a more natural-language approach to searching documentation. You can access this feature either through the **Ask AI** button on the bottom right of the website, or the **Search with AI** button from the search bar. + ### llms.txt -Any AI that searches the website has access to an `llms.txt` file at the site directory root. This file serves as a curated index of content, ensuring it can find relevant information quickly. +The site hosts an `llms.txt` file at the root directory, providing a curated index of content for AI crawlers and tools to find relevant information quickly. -### Context-Optimization +### Context Optimization -Markdown files are the best format to feed an LLM context. Every page on the doc site hosts an `.md` version, which is accessible from the `Copy` dropdown at the top of the page. The full list of copy commands include: +Markdown files are the best format to feed an LLM context. Every page on the doc site hosts an `.md` version, which is accessible from the `Copy` dropdown at the top of the page. The copy options include: -- Copying the contents of the raw `.md` file. -- A link to the `.md` version of the page. -- A button that adds the page URL to a chat window in either `Claude` or `ChatGPT` chat windows. -- A one-click MCP server setup in either `VS Code` or `Cursor`. - - - Every page has a markdown version - -## RAG-Optimized Infrastructure -The site architecture is being optimized for Retrieval-Augmented Generation (RAG). -Semantic Precision -The core pages are undergoing a semantic audit to eliminate ambiguous language and implement self-contained chunking. -When your RAG pipeline retrieves a snippet, it contains all the context necessary for the LLM to understand the logic without needing to crawl adjacent pages. -The llms.txt Standard -The site hosts a specialized llms.txt index. This serves as a high-density "map" for crawlers, ensuring your AI assistants are always referencing the official source of truth. -Overall Developer Experience -We have removed the friction between the documentation and the codebase with features designed for high-stakes environments. -Copy for LLM: Powered by Redocly, our code samples feature a specialized "Copy for LLM" button. This removes UI clutter and formats the code with the metadata headers LLMs need for perfect prompt injection. -Hybrid Tutorials: Our guides are built as "Modular Lego Blocks." They provide the narrative "Why" for human reviewers and the deterministic "How" for AI agents to execute without confusion. +- Copy the contents of the raw `.md` file. +- Copy a link to the `.md` version of the page. +- Open the page in a `Claude` or `ChatGPT` chat window. +- Set up the xrpl.org MCP server in `VS Code` or `Cursor` with one click. From 39f5b9ab6606a1dc8306fd32cf9bc9eea63880e2 Mon Sep 17 00:00:00 2001 From: Oliver Eggert Date: Fri, 10 Apr 2026 16:04:06 -0700 Subject: [PATCH 3/4] fix spelling, grammar, and formatting --- resources/dev-tools/ai-tools.md | 58 ++++++++++++++++++++------------- 1 file changed, 36 insertions(+), 22 deletions(-) diff --git a/resources/dev-tools/ai-tools.md b/resources/dev-tools/ai-tools.md index 7e9941b843..d321513374 100644 --- a/resources/dev-tools/ai-tools.md +++ b/resources/dev-tools/ai-tools.md @@ -4,46 +4,60 @@ seo: --- # AI Tools -AI tools help you build on the XRP Ledger faster by giving AI models access to up-to-date XRPL documentation, SDK references, and code samples. +Augment your AI with additional tools to accelerate development, automate integrations, and deploy secure solutions on the XRP Ledger. ## Model Context Protocol (MCP) Servers -AI models are limited by their training data, which may not include the latest XRPL features or SDK updates. MCP servers solve this by giving your AI real-time access to current documentation through a standardized interface. Instead of relying on potentially outdated training data, your AI can query an MCP server for accurate, up-to-date context. Below is a list of available MCP servers: +AI models are limited by their training data, which will always be out-of-date with the most recent XRPL features and SDK improvements. MCP servers solve this by giving your AI real-time access to current documentation through a standardized interface. Instead of relying on potentially outdated training data, your AI can query an MCP server for accurate, up-to-date context. Below is a list of available MCP servers: -- **Context7**: A large collection of searchable repos and websites. XRPL-relevant info includes: - - [xrpl.org](https://xrpl.org/docs) - - [opensource.ripple.com](https://opensource.ripple.com/) - - [docs.xrplevm.org](https://docs.xrplevm.org/) - - [XRPL JavaScript SDK](https://github.com/xrplf/xrpl.js) - - [XRPL Python SDK](https://github.com/xrplf/xrpl-py) - - [XRPL Go SDK](https://github.com/xrplf/xrpl-go) -- **xrpl.org MCP Server**: Search across documentation on xrpl.org for a given query. +### Context7 + +[Context7](https://context7.com/) is a searchable database of user-submitted repos and websites. Official XRPL docs are maintained on the site and include: +- [xrpl.org](https://xrpl.org/docs): The official developer portal for up-to-date XRPL docs, including use cases, concepts, tutorials, references, and code samples. +- [opensource.ripple.com](https://opensource.ripple.com/): The technical doc site for all features in development by Ripple. +- [docs.xrplevm.org](https://docs.xrplevm.org/): The technical doc site for the XRPL EVM Sidechain. +- [XRPL JavaScript SDK](https://github.com/xrplf/xrpl.js) +- [XRPL Python SDK](https://github.com/xrplf/xrpl-py) +- [XRPL Go SDK](https://github.com/xrplf/xrpl-go) + +To set up Context7, see: [Installation](https://github.com/upstash/context7?tab=readme-ov-file#installation). + +### xrpl.org MCP Server + +The xrpl.org site hosts a standalone MCP server, which only contains documentation hosted on the site. From any doc page, you can click the **Copy** dropdown by the page title and select either **Connect to Cursor** or **Connect to VS Code**. If you are using a different code editor, you can manually configure the MCP server using the `https://xrpl.org/mcp` endpoint. -## SKILLS.md +## SKILL.md -A SKILLS.md file provides behavioral instructions for AI models working with XRPL code. It defines domain-specific rules — like transaction validation, fee handling, and security best practices — so the AI follows correct patterns without you having to explain them each time. +A `SKILL.md` file provides behavioral instructions for AI models working with XRPL code. It defines specific steps and rules to produce more precise outcomes, such as forming transactions, implementing security best practices, or issuing tokens. By loading a skill in your coding agent, you reduce the need for verbose or repeated queries. -- **[XRPL Development Skill for Claude Code](https://github.com/XRPL-Commons/xrpl-dev-skills)**: A comprehensive Claude Code skill for modern XRP Ledger development, provided by XRPL Commons. This skill uses Claude Code's progressive disclosure pattern. The main `SKILL.md` provides core guidance, and Claude reads specialized markdown files only when needed for specific tasks. -- **[Generate Release Notes](https://github.com/XRPLF/xrpl-dev-portal/blob/master/.claude/skills/generate-release-notes/SKILL.md)**: This skill generates a draft release notes blog post for new versions of `rippled`. This is intended for those contributing to the XRPL documentation site. +### XRPL Development Skill for Claude Code + +A comprehensive Claude Code skill for modern XRP Ledger development, provided by XRPL Commons. This skill uses Claude Code's progressive disclosure pattern; the main `SKILL.md` provides core guidance, and Claude reads specialized markdown files only when needed for specific tasks. For a full list of available skills, as well as installation instructions, check the [GitHub repo](https://github.com/XRPL-Commons/xrpl-dev-skills?tab=readme-ov-file#xrpl-development-skill-for-claude-code). + +### Generate Release Notes + +The `generate-release-notes` skill generates a draft release notes to publish to the xrpl.org blog. This skill is intended for contributors to the XRPL docs. For additional details, check the [GitHub repo](https://github.com/XRPLF/xrpl-dev-portal/blob/master/.claude/skills/generate-release-notes/SKILL.md). ## Site Optimizations -### AI Search +The XRPL developer portal has been updated with AI optimizations to serve both human and AI audiences. -We include an AI chatbot on xrpl.org, opensource.ripple.com, and docs.xrplevm.org to provide a more natural-language approach to searching documentation. You can access this feature either through the **Ask AI** button on the bottom right of the website, or the **Search with AI** button from the search bar. +### AI Chatbot + +An AI chatbot is hosted on [xrpl.org](https://xrpl.org/docs), [opensource.ripple.com](https://opensource.ripple.com/), and [docs.xrplevm.org](https://docs.xrplevm.org/) for users who prefer a more natural-language approach to documentation. You can access the chatbots from the **Ask AI** button at the bottom right of each website, or from the **Search with AI** button embedded in the search bar. ### llms.txt -The site hosts an `llms.txt` file at the root directory, providing a curated index of content for AI crawlers and tools to find relevant information quickly. +The site hosts an [llms.txt](https://xrpl.org/llms.txt) file at the root directory, providing a curated index of content for AI crawlers and tools to find relevant information quickly. If you don't want to use the MCP servers, you can provide this file to your AI as context. ### Context Optimization -Markdown files are the best format to feed an LLM context. Every page on the doc site hosts an `.md` version, which is accessible from the `Copy` dropdown at the top of the page. The copy options include: +Markdown is an efficient format for providing documentation context to an AI. Every page on the doc site hosts an `.md` version, which is accessible from the **Copy** dropdown at the top of the page. The dropdown includes options to: -- Copy the contents of the raw `.md` file. -- Copy a link to the `.md` version of the page. -- Open the page in a `Claude` or `ChatGPT` chat window. -- Set up the xrpl.org MCP server in `VS Code` or `Cursor` with one click. +- Copy the contents of the `.md` file. +- View the page as an `.md` file. +- Open **Claude** or **ChatGPT** with the contents of the page automatically loaded as context. +- Set up the xrpl.org MCP server in **VS Code** or **Cursor** with one click. From 9f7c6755174f28c1d818af2ce460d05eb284886f Mon Sep 17 00:00:00 2001 From: oeggert <117319296+oeggert@users.noreply.github.com> Date: Fri, 24 Apr 2026 11:39:04 -0700 Subject: [PATCH 4/4] Update resources/dev-tools/ai-tools.md Co-authored-by: Amarantha Kulkarni --- resources/dev-tools/ai-tools.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/resources/dev-tools/ai-tools.md b/resources/dev-tools/ai-tools.md index d321513374..554ce6ae54 100644 --- a/resources/dev-tools/ai-tools.md +++ b/resources/dev-tools/ai-tools.md @@ -55,7 +55,7 @@ The site hosts an [llms.txt](https://xrpl.org/llms.txt) file at the root directo ### Context Optimization -Markdown is an efficient format for providing documentation context to an AI. Every page on the doc site hosts an `.md` version, which is accessible from the **Copy** dropdown at the top of the page. The dropdown includes options to: +Markdown is an efficient format for providing documentation context to an AI. Every documentation page on the site hosts an `.md` version, which is accessible from the **Copy** dropdown at the top of the page. The dropdown includes options to: - Copy the contents of the `.md` file. - View the page as an `.md` file.