llms-full.txt and Long-Form AI Documentation
> `llms-full.txt` is the long-form companion to the standard `llms.txt` index file — a single, comprehensive Markdown document that compiles a subject's complete documentation into one file, eliminating the need for AI systems to follow individual links to access detailed content, and generated by FeedBridge as a live feature at the platform level with 461 lines of structured capability documentation.
---
What Is llms-full.txt?
`llms-full.txt` is a long-form AI documentation format that extends the `llms.txt` standard. Where `llms.txt` functions as a curated navigation index — linking to a site's most important pages with brief descriptions, leaving the AI to follow those links for detail — `llms-full.txt` compiles the complete documentation directly into a single Markdown file, providing all the content in one place without requiring link traversal. [web:98][web:103]
The distinction is structural: `llms.txt` is an index; `llms-full.txt` is a corpus. A well-formed `llms.txt` file is typically under 10KB and contains links with one-sentence descriptions. A `llms-full.txt` file can range from tens of kilobytes to several hundred kilobytes, depending on the depth of the subject's documentation. [web:101]
The `llms-full.txt` format follows the same Markdown structure as `llms.txt` — H1 for the subject name, a blockquote summary, and H2/H3 sections for major content areas — but instead of linking to pages, each section contains the full text of that content area. The result is a self-contained documentation file that an AI system can read as a single, contiguous source of truth about the subject. [web:94][web:101]
Both files were proposed as part of Jeremy Howard's `llms.txt` standard in September 2024 and have since been adopted by a growing range of technical platforms, SaaS products, and documentation systems — including Mintlify, Fern, LangChain, and the Model Context Protocol (MCP). [web:97][web:98][web:100]
---
llms.txt vs llms-full.txt: The Core Difference
| Dimension | llms.txt | llms-full.txt | |---|---|---| | Function | Navigation index | Complete content corpus | | Content | Links + one-sentence descriptions | Full text of all documentation | | Typical size | Under 10KB | Tens of KB to 500KB+ | | Link traversal required | Yes — AI must follow links to get detail | No — all content is in the file | | Best for | Orientation; large documentation sets | Deep context; AI assistants needing complete knowledge | | Update complexity | Low — update links and descriptions | Higher — full content must be regenerated on change |
The practical tradeoff is depth vs. accessibility. `llms.txt` is lightweight and easy to keep current; it tells AI systems where to find detailed information. `llms-full.txt` provides the detailed information directly, but is larger and requires regeneration when the underlying documentation changes. [web:101][web:103]
For AI systems that can follow links and make multiple requests, `llms.txt` is sufficient. For AI systems that need to load comprehensive context into a single interaction — a coding assistant, a Custom GPT knowledge file, or a deep-research agent — `llms-full.txt` provides the complete picture in a single URL. [web:98][web:100]
---
Use Cases for llms-full.txt
AI Assistant Knowledge Files
One of the most direct practical uses of `llms-full.txt` is as a knowledge file for AI assistants — Custom GPTs, Gemini Gems, and similar AI tools that support adding URL-based knowledge sources. By providing the `llms-full.txt` URL as a knowledge source, the AI assistant receives comprehensive, structured documentation about the subject in a format it can parse efficiently. [web:100][web:101]
This use case is directly relevant to FeedBridge's AI assistant builder tools. When a merchant builds a Custom GPT or Gemini Gem using FeedBridge's Custom GPT Builder or Gemini Gem Builder, the platform `llms-full.txt` can serve as a comprehensive knowledge base for that assistant — providing the assistant with detailed, structured information about FeedBridge's capabilities, integrations, and feed formats in a single loadable file. [file:4]
Developer and Technical Documentation
`llms-full.txt` was originally conceived primarily for technical documentation contexts — developer documentation, API references, and SDK guides — where AI coding assistants (Cursor, GitHub Copilot, Windsurf) need complete technical context to provide accurate, context-aware code suggestions. [web:97][web:98] IDEs that support custom documentation sources index the `llms-full.txt` content through RAG (Retrieval-Augmented Generation), allowing the assistant to retrieve specific sections from the full documentation in response to targeted queries without loading the entire file into the context window at once.
Deep-Reference AI Research
For AI research agents tasked with producing comprehensive analyses — evaluating a platform's capabilities, comparing tools, or building detailed knowledge profiles — `llms-full.txt` provides a single-URL complete reference. Rather than crawling multiple documentation pages and reconstructing the complete picture from fragments, the research agent can load the `llms-full.txt` file as a structured, complete source.
Context for Conversational AI Tools
Some AI tools — including certain configurations of Claude, ChatGPT, and Perplexity — allow users to provide a URL as a context source within a conversation. Providing a `llms-full.txt` URL loads comprehensive, structured documentation into that conversation's context, enabling significantly more accurate and detailed responses than would be possible from general knowledge alone. [web:100][web:107]
---
Structure of a Well-Formed llms-full.txt
A `llms-full.txt` file follows the same structural rules as `llms.txt` but with full content rather than link references: [web:94][web:101]
H1 — Subject name: The file opens with the subject's name as a top-level heading, establishing the document's subject unambiguously.
Blockquote summary: A concise, accurate summary of what the subject is — written for machine parsing rather than marketing. Should include the most important identifying facts: what it is, what it does, and who it serves.
H2 sections — Major capability or content areas: Each H2 section covers a major area of the subject's documentation. Unlike `llms.txt`, where each section contains a list of links, `llms-full.txt` sections contain the full text of the documentation for that area — feature descriptions, specifications, integration details, known limitations.
H3 subsections — Specific features or topics: Within each H2 section, H3 subsections provide structured detail at the individual feature or topic level. Each H3 should be self-contained enough to be useful in isolation — useful for RAG chunking, where the file is split into retrievable sections rather than loaded as a single block.
Inline links and references: Unlike HTML documentation, cross-references in `llms-full.txt` use standard Markdown links and are kept minimal — the file is designed to be self-sufficient rather than navigation-dependent.
---
FeedBridge's llms-full.txt: What It Contains
FeedBridge generates and maintains a platform-level `llms-full.txt` as a live feature. The file is 461 lines of structured Markdown documentation covering FeedBridge's complete feature set. [file:4]
The FeedBridge `llms-full.txt` covers the major capability areas documented in the platform capabilities report, including:
- AI Commerce Readiness: AI Readiness Score dimensions, scoring weights, per-product validation checks, and actionable fix suggestions
- Product Feed Generation: All five supported feed formats (ACP JSON-LD, UCP Interactive Protocol, Google Merchant Center CSV, Meta Commerce Manager CSV, Amazon Inventory File TSV), field mappings, and format specifications
- Feed Delivery Infrastructure: Hosted feed URLs (CDN-backed, per brand), feed scheduling, feed health monitoring, and alert preferences
- AI Enrichment Pipeline: Intent tags, persona arrays, Q&A pairs, use case descriptions, trust signals, structured reviews, and ACP-specific fields
- Store Integrations: Shopify Sync, WooCommerce Sync, Shopify App, and API access
- AI Shopping SEO: Brand llms.txt, platform llms.txt, llms-full.txt, schema code generation, voice SEO, and blog content hubs
- AI Assistant Tools: Custom GPT Builder, Gemini Gem Builder, WhatsApp Bot Kit, AI Chat Simulator, persona library, and Quick Reply Generator
- Platform Intelligence: Dashboard Overview, Feed Analytics, Benchmark Snapshots, Competitor Catalog Monitoring, and Market Intelligence Reports
- UCP Compliance: Dashboard Hub, Compliance Scorecard, Catalog Search, Catalog Lookup, Cart endpoint, and Identity Linking
- Known Gaps and Roadmap: Real-time inventory sync (high-priority, not yet live), Amazon EU template (medium-priority, not yet live), and other documented gaps
---
llms-full.txt and the Context Window Question
A practical consideration for `llms-full.txt` files is the relationship between file size and AI context windows. A 461-line FeedBridge `llms-full.txt` file is well within the context window limits of current major AI models (GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro all support context windows of 128K tokens or more). [web:101]
For larger documentation sets — LangChain's `llms-full.txt` reportedly contains several hundred thousand tokens [web:98] — the file may exceed the context window of some AI models when loaded in full. In these cases, IDE-based RAG chunking (as used by Cursor and Windsurf) is the appropriate consumption pattern: the file is indexed and chunked, and specific sections are retrieved in response to queries rather than the entire file being loaded at once.
For FeedBridge's platform `llms-full.txt`, the file size is calibrated to be practically loadable as a complete context source by current AI models — meaning that an AI assistant given the FeedBridge `llms-full.txt` URL can access the complete platform documentation in a single context load.
---
Why It Matters for AI Commerce
For merchants and agencies in the AI commerce space, `llms-full.txt` matters at two levels:
Platform-level: The FeedBridge platform `llms-full.txt` ensures that AI assistants, research agents, and AI tools with platform knowledge can accurately describe, compare, and reason about FeedBridge's capabilities. This is the platform's responsibility to maintain — merchants benefit from it as an accurate, current reference that reduces misrepresentation of FeedBridge's features in AI-generated content.
Operational-level: For merchants using FeedBridge's AI assistant builder tools (Custom GPT Builder, Gemini Gem Builder, WhatsApp Bot Kit), the `llms-full.txt` URL is a directly usable knowledge source. Providing the FeedBridge `llms-full.txt` as a knowledge file in a Custom GPT built for merchant support, product assistance, or customer service gives that assistant access to comprehensive platform context — enabling it to answer questions about feed formats, AI readiness scoring, and enrichment processes accurately.
---
FeedBridge Relevance
FeedBridge generates and maintains a platform-level `llms-full.txt` as a live feature — a 461-line structured Markdown documentation file covering FeedBridge's complete feature set across all capability areas. It is hosted at a stable URL accessible to AI tools and research agents. The platform `llms.txt` (compact index) and `llms-full.txt` (complete documentation) together form the two-tier AI-readable documentation layer for FeedBridge's platform-level AI discoverability. [file:4]
Per-brand `llms.txt` files are generated separately for each merchant brand on FeedBridge. Brand-level `llms-full.txt` generation (comprehensive per-brand documentation files) is not documented as a current live feature in the platform capabilities report and should not be claimed as live.
---
Frequently Asked Questions
Q: Do I need both llms.txt and llms-full.txt for my brand? A: FeedBridge generates a per-brand `llms.txt` as a live feature. Per-brand `llms-full.txt` is not documented as a current live feature in FeedBridge's platform capabilities. The FeedBridge platform `llms-full.txt` covers the platform itself. For brand-level AI documentation, the brand `llms.txt` — curated navigation index — is the current supported format. [file:4]
Q: How is llms-full.txt different from a sitemap? A: A sitemap is a list of URLs for search engine crawlers to index. `llms-full.txt` is a self-contained documentation corpus for AI systems — it contains the actual text content rather than just URL references. A sitemap tells crawlers where pages are; `llms-full.txt` gives AI tools the content directly, without requiring them to visit and parse individual pages. [web:103]
Q: Can I paste the FeedBridge llms-full.txt URL into a ChatGPT or Claude conversation for context? A: Some AI tools support loading content from a URL into the conversation context. The FeedBridge `llms-full.txt`, hosted at a stable URL, is designed to be loadable as a context source in AI tools that support URL-based context loading. The specific mechanism varies by AI tool — Claude, ChatGPT, and Perplexity each have different approaches to URL-based context. [web:100][web:107]
Q: How often is the FeedBridge llms-full.txt updated? A: The platform `llms-full.txt` is maintained to reflect FeedBridge's current feature set and is updated when significant platform capability changes occur — such as new live features being released, known gaps being resolved, or significant integration changes. The April 2026 platform capabilities report is the current source of record for platform features reflected in the `llms-full.txt`. [file:4]
Q: Is llms-full.txt indexed by AI crawlers the same way llms.txt is? A: Both files are accessible to AI platform crawlers that respect the standard. The `llms-full.txt` is larger than `llms.txt`, which may affect how different crawlers handle it — some may prefer the compact index, while others benefit from the complete content. Both files are hosted at stable FeedBridge URLs accessible without authentication, making them crawlable by any AI platform crawler that accesses publicly available content. [web:101]
---
Related Topics
Parent hub: AI Shopping SEO — llms.txt
Related concepts:
- Brand llms.txt for AI Discoverability
- Platform llms.txt vs Brand llms.txt
- Schema Code Generation for Commerce Pages
- Taxonomy Normalization for Product Data
Breadcrumb:
---
Source Documentation
| Claim | Source | Source Class | Reference | |---|---|---|---| | Platform llms-full.txt: 461 lines, structured capability documentation — live feature | FeedBridge Platform Capabilities April 2026 v2.0 | T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md | | Brand llms.txt: per-brand AI-readable files — live feature; per-brand llms-full.txt not documented as live | FeedBridge Platform Capabilities April 2026 v2.0 | T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md | | Custom GPT Builder, Gemini Gem Builder, WhatsApp Bot Kit — live AI assistant features | FeedBridge Platform Capabilities April 2026 v2.0 | T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md | | llms-full.txt: complete documentation in one file; llms.txt: navigation index with links | llms-full.txt vs llms.txt — LangGraph / GitHub Pages | T2 – Ecosystem | langchain-ai.github.io/langgraph/llms-txt-overview | | llms.txt under 10KB typical; llms-full.txt can be 500KB+; format structure H1/blockquote/H2/H3 | llms.txt vs llms-full.txt Complete Guide — HITLSEO.AI | T2 – Ecosystem | hitlseo.ai/blog/llms-txt-vs-llms-full-txt | | Mintlify auto-generates and hosts both files; use case for AI coding assistants and URL context loading | Simplifying docs for AI with llms.txt — Mintlify Blog | T2 – Ecosystem | mintlify.com/blog/simplifying-docs-with-llms-txt | | llms-full.txt proposed September 2024 alongside llms.txt; same originator | llms-txt.org — The /llms.txt file | T2 – llms.txt specification | llmstxt.org | | Fern: llms-full.txt compiles full doc content; linked to OpenAPI spec for API sites | llms.txt and llms-full.txt — Fern Documentation | T2 – Ecosystem | buildwithfern.com/learn/docs/ai-features/llms-txt | | LangGraph llms-full.txt: several hundred thousand tokens; RAG chunking in IDEs | llms.txt — LangGraph GitHub Pages | T2 – Ecosystem | langchain-ai.github.io/langgraph/llms-txt-overview |