Add auto-generated llms.txt + AGENTS.md, CHANGELOG.md, and bump to 1.1.0

[ErickXavier]

What

Makes the directory readable by LLMs and AI coding agents, and adds release hygiene:

• public/llms.txt — served at https://nosignups.net/llms.txt following the llmstxt.org convention. Since the site is a React SPA, an LLM fetching the root URL gets an empty HTML shell; this file summarizes the project and embeds the complete tool directory (all 210 tools grouped by category, with links, descriptions, source repos, and licenses), so LLMs can read the whole list in one fetch.
• AGENTS.md — repo-root instructions for AI coding agents per the agents.md convention: commands, project layout, tools.json schema, contribution rules, plus a compact auto-generated tool index so agents can check for duplicates before adding a tool.
• scripts/generateLlmsDocs.mjs — zero-dependency Node script that regenerates the tool lists in both files from tools.json (block between BEGIN/END GENERATED TOOL LIST markers; hand-written prose is untouched). Sorting mirrors the site: featured first, then stars. Runs automatically as the first step of npm run build, so the published files always reflect the latest tools.json; also runnable standalone via npm run generate:llms.
• vite.config.ts — a small build-only plugin that copies AGENTS.md into dist/ so it is also served at https://nosignups.net/AGENTS.md (linked from llms.txt).
• CHANGELOG.md — Keep a Changelog format with SemVer versioning; 1.1.0 covers this PR, 1.0.0 the initial release. Routine tools.json edits are explicitly out of its scope.
• package.json — version bumped 1.0.0 → 1.1.0 (additive, backwards-compatible changes).

Why

More and more people find tools by asking an LLM. These files let assistants discover and correctly cite the NoSignups directory — including every listed tool — and help coding agents contribute to the repo following your guidelines.

Verification

• npm run build passes; dist/ contains both llms.txt and AGENTS.md with all 210 tools embedded.
• Generation is idempotent (running it twice produces byte-identical files).
• No runtime code paths touched — the vite change is build-time only (apply: "build"), so npm run dev is unaffected.

[BraveOPotato]

Hi,

Thank you for the PR!

As far as I understand, AGENTS.md is a file used to tell LLM agents how to work on the code base, but I'm not exactly sure it needs to be served at /AGENTS.md from the browser. Would you mind elaborating on that?

[ErickXavier]

Hi,

Thank you for the PR!

As far as I understand, AGENTS.md is a file used to tell LLM agents how to work on the code base, but I'm not exactly sure it needs to be served at /AGENTS.md from the browser. Would you mind elaborating on that?

That's not for the browser, the AGENTS.md, as my PR describes:

AGENTS.md — repo-root instructions for AI coding agents per the agents.md convention: commands, project layout, tools.json schema, contribution rules, plus a compact auto-generated tool index so agents can check for duplicates before adding a tool.

[BraveOPotato]

I saw that but the reason why I asked is because vite.config.ts has a custom plugin you wrote which copies the AGENTS.md file to the /dist directory.

Since deployment takes everything in the /dist directory and uploads it to static hosting, it makes the AGENTS.md file accessible at /AGENTS.md route.

I'm essentially wondering why the custom plugin is necessary if we don't want the file to be accessible from the browser.

[ErickXavier]

I saw that but the reason why I asked is because vite.config.ts has a custom plugin you wrote which copies the AGENTS.md file to the /dist directory.

Since deployment takes everything in the /dist directory and uploads it to static hosting, it makes the AGENTS.md file accessible at /AGENTS.md route.

I'm essentially wondering why the custom plugin is necessary if we don't want the file to be accessible from the browser.

So other dev's agents will have the context and how to work on FckSignups :)

[BraveOPotato]

But the file would be committed to the repo anyways. So in theory, when they want to work on it, they'd have the repo cloned, which means they'll have the file already without navigating to /AGENTS.md in the browser. I don't see why it needs to be served via static hosting.

[ErickXavier]

But the file would be committed to the repo anyways. So in theory, when they want to work on it, they'd have the repo cloned, which means they'll have the file already without navigating to /AGENTS.md in the browser. I don't see why it needs to be served via static hosting.

I fed it to the llms.txt also, so the website will already introduce the file to the agents, whenever they visit the site before they visit the repo.