Autonomi LLM Documentation

Machine-readable documentation for the Autonomi Network, designed to help AI assistants understand and explain the platform to developers.

What This Is

This repository contains:

• llms.txt - Main index file following the llms.txt standard
• Markdown documentation - Clean, comprehensive content about Autonomi's architecture, features, and developer experience
• Page variants - .md versions of key autonomi.com pages for AI consumption
• Cloudflare Worker config - Source-controlled Worker code/config in worker/, with manual deployment workflows documented in the Worker runbook

How It Works

Content in this repo is served at autonomi.com via Cloudflare Worker.

Files appear to live on autonomi.com but are actually served from this GitHub repo, giving us:

• Version control for all LLM documentation
• Easy collaborative editing
• Single source of truth
• Clean content without HTML wrappers

Public Serving Policy

The Cloudflare Worker serves .md, /llms.txt, and /llms-full.txt from this repo at canonical autonomi.com URLs, except for internal repository prefixes.

Public by default:

• .md files outside internal prefixes
• /llms.txt
• /llms-full.txt

Internal prefixes not served from GitHub raw content:

• /worker/
• /.github/

To publish machine-readable content, add a .md file outside the internal prefixes. To keep operational or repository-internal documentation out of canonical autonomi.com serving, place it under an internal prefix or use a non-served extension. See ADR-0006 for the decision record.

Publishing and Managing Content

This repo is for machine-readable, LLM-parsable, AI-optimised content served from the autonomi.com domain.

It provides Markdown/plain-text versions of key autonomi.com pages, and any other content the team wants available at stable autonomi.com URLs. These files are maintained in GitHub in parallel with the human-readable HTML pages managed in Framer.

For pages that have a Markdown counterpart in this repo, the .md URL is the LLM-friendly version of the page. For example, a human-facing page on autonomi.com can have a Markdown counterpart such as https://autonomi.com/overview.md when overview.md exists in this repository.

This repo also provides:

• llms.txt: a concise index for AI assistants and crawlers
• llms-full.txt: a fuller context file for AI assistants that want more inline detail
• Markdown versions of selected public pages and reference material
• foundational whitepapers in both Markdown and PDF form
• source-controlled Worker code/config for serving the content at autonomi.com

For a normal content update:

1. Edit the existing public content file, or add a new public .md file outside internal prefixes.
2. Update llms.txt so it includes every public content URL that should be discoverable.
3. Update llms-full.txt so it reflects the current public content set and messaging.
4. Open a PR.
5. Review the Markdown and the final autonomi.com URL it will map to.
6. Merge the PR to main.
7. Check the live URL after merge if needed.

After a content PR is merged to main, the Cloudflare Worker serves the updated content automatically from GitHub raw content. There is usually no separate deployment step for content-only changes. Served documentation responses use Cache-Control: public, max-age=300, so changes may take up to about five minutes to appear at autonomi.com.

When public content is added, removed, renamed, or substantially changed:

• add new public URLs to llms.txt
• remove deleted public URLs from llms.txt
• update llms-full.txt to match the current content and messaging
• check whether the content should be linked from an existing section or whether a new section is needed
• keep URLs stable where possible, because agents and external tools may cache or reference them

Public content is served when it matches one of these rules:

• .md files outside internal prefixes
• /llms.txt
• /llms-full.txt

Examples:

• overview.md is served at https://autonomi.com/overview.md
• whitepapers/autonomous-network.md is served at https://autonomi.com/whitepapers/autonomous-network.md
• llms.txt is served at https://autonomi.com/llms.txt
• llms-full.txt is served at https://autonomi.com/llms-full.txt

Whitepapers

Foundational whitepapers are kept in two forms:

• Markdown versions for LLM-readable content served from autonomi.com
• PDF versions for stable, shareable, human-readable original documents

When adding or updating a whitepaper:

1. Keep the Markdown version under whitepapers/.
2. Keep the corresponding PDF in the same directory when a PDF counterpart exists.
3. For whitepapers that have a PDF counterpart, link from the Markdown page to the PDF using an explicit GitHub raw URL.
4. Add or update the Markdown URL in llms.txt.
5. Update llms-full.txt so the whitepaper set and summary remain current.

The Worker does not currently serve .pdf files from autonomi.com. Whitepaper PDF links should use explicit GitHub raw URLs, for example:

https://raw.githubusercontent.com/WithAutonomi/autonomi-llm-docs/main/whitepapers/Autonomous-Network.pdf

Do not use relative PDF links such as Autonomous-Network.pdf from a served Markdown page, because that would resolve against autonomi.com and fall through to Framer rather than serving the file from this repo.

Under the current policy, any .md file outside internal prefixes can be served from autonomi.com. This includes repo-level and governance Markdown such as README.md and docs/adr/*.md; those files are intentionally visible as public repository context.

Operational or maintenance material that should not be published as canonical Autonomi documentation should live under an internal prefix, or use a non-served extension such as .markdown.

Internal prefixes not served from GitHub raw content:

• /worker/
• /.github/

Use internal locations for:

• Worker source and runbooks
• GitHub workflow files or workflow documentation
• repository-maintenance notes
• operational material that is useful in GitHub but should not appear as public autonomi.com documentation

Changing content is different from changing serving behaviour.

Content-only changes normally need only a PR merge. Worker or serving-policy changes need additional review and, after merge, a manual Worker deployment. For Worker changes, follow the worker/README.markdown runbook: deploy preview, approve the required environment review, smoke test the workers.dev URL, deploy production, approve the required environment review, and smoke test autonomi.com. Treat these as serving changes:

• editing worker/src/index.js
• changing Wrangler config
• changing deploy workflows
• adding new public file types beyond .md, /llms.txt, or /llms-full.txt
• serving PDFs from autonomi.com instead of GitHub raw URLs
• changing internal prefixes or routing policy
• changing cache behaviour or response headers

Serving-policy changes should be reflected in Worker tests and checked against ADR-0006. PDF-linking behaviour is documented in ADR-0004 and should be revisited if the Worker is changed to serve PDFs directly from autonomi.com.

Who This Is For

Primary audience: AI assistants (Claude, ChatGPT, etc.) helping developers build on Autonomi

Secondary audience: Developers who want clean, comprehensive technical documentation

Contributing

This repo is maintained by the Autonomi team. Content should be:

• Written in clear, accessible markdown
• Comprehensive enough for AI assistants to guide developers and inform those wishing to understand the technology
• Free of HTML or formatting that would confuse machine readers

Related

• Autonomi Website
• Developer Documentation
• llms.txt Standard