feat: add SEO, AEO and GEO standards

SEO:
- Web app manifest (manifest.json) for PWA/mobile discoverability

AEO (Answer Engine Optimization):
- Inline SoftwareApplication + Organization + WebSite JSON-LD in index.html
- Inline FAQPage JSON-LD with 11 questions matching the on-page FAQ
- HowTo schema for the three-step install process
- All three were in separate files but never linked from the HTML

GEO (Generative Engine Optimization):
- llms-full.txt with detailed tool catalogue, architecture, FAQ
- llms.txt updated with link to llms-full.txt
- robots.txt references both llms.txt and llms-full.txt
- link rel=help in head pointing to llms.txt

Author: reddy-sh

site/index.html

@@ -8,6 +8,7 @@
     <link rel="canonical" href="https://tokenhud.com/" />
     <meta name="author" content="TokenHUD" />
     <meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1" />
+    <!-- Open Graph -->
     <meta property="og:type" content="website" />
     <meta property="og:site_name" content="TokenHUD" />
     <meta property="og:url" content="https://tokenhud.com/" />
@@ -18,13 +19,212 @@
     <meta property="og:image:height" content="630" />
     <meta property="og:image:alt" content="The TokenHUD dashboard showing token spend, session metering and five-hour plan usage." />
     <meta property="og:locale" content="en_US" />
+    <!-- Twitter -->
     <meta name="twitter:card" content="summary_large_image" />
     <meta name="twitter:title" content="TokenHUD — a heads-up display for the AI agents on your machine" />
     <meta name="twitter:description" content="Live token spend, session metering and plan limits for Claude Code and Codex CLI. Metrics leave, content never does." />
     <meta name="twitter:image" content="https://tokenhud.com/og.png" />
+    <!-- Icons & theme -->
     <meta name="theme-color" content="#1a120c" />
     <link rel="icon" href="/favicon.svg" type="image/svg+xml" />
     <link rel="apple-touch-icon" href="/apple-touch-icon.png" />
+    <link rel="manifest" href="/manifest.json" />
+    <!-- GEO: llms.txt for AI crawlers -->
+    <link rel="help" type="text/plain" href="/llms.txt" title="LLM-readable site summary" />
+    <!-- AEO: structured data (SoftwareApplication + Organization + WebSite) -->
+    <script type="application/ld+json">
+    {
+      "@context": "https://schema.org",
+      "@graph": [
+        {
+          "@type": "SoftwareApplication",
+          "@id": "https://tokenhud.com/#app",
+          "name": "TokenHUD",
+          "alternateName": "Token HUD",
+          "description": "A heads-up display for the AI coding agents running on your machine. Live token metering, estimated spend, and plan-limit windows for Claude Code and Codex CLI — running entirely on your own computer.",
+          "applicationCategory": "DeveloperApplication",
+          "applicationSubCategory": "Developer Monitoring Tool",
+          "operatingSystem": "macOS, Linux",
+          "url": "https://tokenhud.com/",
+          "downloadUrl": "https://github.com/reddy-sh/tokenhud",
+          "codeRepository": "https://github.com/reddy-sh/tokenhud",
+          "programmingLanguage": "Rust",
+          "license": "https://opensource.org/licenses/MIT",
+          "isAccessibleForFree": true,
+          "offers": [
+            {
+              "@type": "Offer",
+              "name": "Solo",
+              "price": "0",
+              "priceCurrency": "USD",
+              "description": "Free forever for one developer on one machine. Unlimited metering, no sampling, full local history. Runs with the network off."
+            },
+            {
+              "@type": "Offer",
+              "name": "Team",
+              "priceCurrency": "USD",
+              "description": "Per seat. Cross-machine rollups, shared budgets, retained history, chargeback reports, and alerts to Slack and email. Planned — not yet available.",
+              "availability": "https://schema.org/PreOrder"
+            }
+          ],
+          "featureList": [
+            "Discovers AI coding agents and their sessions on the host with no configuration",
+            "Meters token usage and estimated spend by session, agent, model and project",
+            "Shows your plan's real five-hour and seven-day usage windows and when they reset",
+            "Reports which agents finished while you were away",
+            "Never reads prompt text, completion text, or source code"
+          ],
+          "softwareRequirements": "Rust toolchain to build from source. Two static binaries; no interpreter, no package manager.",
+          "author": { "@id": "https://tokenhud.com/#org" },
+          "publisher": { "@id": "https://tokenhud.com/#org" }
+        },
+        {
+          "@type": "Organization",
+          "@id": "https://tokenhud.com/#org",
+          "name": "TokenHUD",
+          "url": "https://tokenhud.com/",
+          "logo": "https://tokenhud.com/logo.png",
+          "sameAs": ["https://github.com/reddy-sh/tokenhud"]
+        },
+        {
+          "@type": "WebSite",
+          "@id": "https://tokenhud.com/#website",
+          "url": "https://tokenhud.com/",
+          "name": "TokenHUD",
+          "publisher": { "@id": "https://tokenhud.com/#org" }
+        }
+      ]
+    }
+    </script>
+    <!-- AEO: FAQ structured data -->
+    <script type="application/ld+json">
+    {
+      "@context": "https://schema.org",
+      "@type": "FAQPage",
+      "@id": "https://tokenhud.com/#faq",
+      "mainEntity": [
+        {
+          "@type": "Question",
+          "name": "How do I see how many tokens Claude Code is using?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "TokenHUD meters Claude Code's token usage live on your own machine. It reads Claude Code's local session transcripts and reports input, output and cache tokens per session, model, and project, plus your plan's five-hour and seven-day usage windows and when they reset."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Does TokenHUD see my prompts or my code?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "No. Prompt text, completion text, source code and tool-call arguments are never read into the payload at all — not filtered out afterwards. The transcript parser reads only the numeric usage fields from assistant records. There is no code path that collects content, so no configuration mistake can expose it."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Is TokenHUD free?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Yes, free forever for one developer on one machine, with unlimited metering and no sampling. A paid per-seat team tier adds cross-machine rollups, shared budgets, retained history and chargeback reports. The free tier is not degraded to drive upgrades."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Does TokenHUD work offline?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Yes. The free tier runs entirely on your machine and sends nothing anywhere. The agent, the server and the dashboard all run on localhost, so it works with the network off."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Will TokenHUD slow down my machine?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "No. A collection cycle takes about 0.06 seconds in steady state. The agent indexes a transcript corpus incrementally by byte offset, so after the first pass it reads only what was appended."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Which AI coding agents does TokenHUD support?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Claude Code and Codex CLI today. TokenHUD also detects Cursor, Gemini CLI, Windsurf, Devin and other assistants installed on the machine and lists them, marking the ones no collector reads yet."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "How do I track the cost of my AI coding agents?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "TokenHUD prices your measured token usage at published API list rates and shows estimated value per session, per model and per day. On a subscription plan the CLI reports $0 per request because the plan is a flat fee, so the figure is labelled an estimate."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "When does my Claude usage limit reset?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "TokenHUD reads the real five-hour and seven-day window state that Claude Code caches locally and shows the percentage used with a live countdown to each reset."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Can I monitor MCP servers?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "MCP server discovery and health is on the roadmap and not yet built. TokenHUD today discovers and reports Claude Code agent processes on the host."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "What does TokenHUD install?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Two static binaries, about 4 MB together. No interpreter, no npx, no pip, no package manager. The dashboard is a single self-contained HTML file with no CDN reference."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Is TokenHUD open source?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Yes, MIT licensed, at github.com/reddy-sh/tokenhud. The agent, server and dashboard are all in the repository and it ships a self-test that runs the real collectors against your machine with nothing mocked."
+          }
+        }
+      ]
+    }
+    </script>
+    <!-- AEO: HowTo schema for installation -->
+    <script type="application/ld+json">
+    {
+      "@context": "https://schema.org",
+      "@type": "HowTo",
+      "@id": "https://tokenhud.com/#howto-install",
+      "name": "How to install TokenHUD",
+      "description": "Install the TokenHUD agent on macOS or Linux to start metering your AI coding agents' token usage.",
+      "totalTime": "PT2M",
+      "step": [
+        {
+          "@type": "HowToStep",
+          "position": 1,
+          "name": "Run the install script",
+          "text": "Open a terminal and run: curl -fsSL https://tokenhud.com/install.sh | sh"
+        },
+        {
+          "@type": "HowToStep",
+          "position": 2,
+          "name": "The agent starts automatically",
+          "text": "The install script downloads the agent binary, installs a login service, and starts the agent. It begins discovering AI coding agents on your machine immediately."
+        },
+        {
+          "@type": "HowToStep",
+          "position": 3,
+          "name": "Open the dashboard",
+          "text": "Sign in at platform.tokenhud.com or open your self-hosted board at localhost:9746 to see live token usage, spend, and plan limits."
+        }
+      ]
+    }
+    </script>
   </head>
   <body>
     <div id="root"></div>

site/public/llms-full.txt

@@ -0,0 +1,131 @@
+# TokenHUD — full context for language models
+
+> A heads-up display for the AI agents running on your machine. Live token
+> metering, estimated spend, and plan-limit windows for Claude Code, Codex CLI,
+> GitHub Copilot CLI, Devin CLI, and 22 other tools — running entirely on the
+> developer's own computer. Free and open source (MIT).
+
+## What it does
+
+TokenHUD watches a developer's own machine and reports what the AI coding
+agents on it are doing:
+
+- Which sessions are running right now and how long they have been open
+- How many tokens each session has consumed (input, output, cache read, cache write)
+- What those tokens are worth at published API list prices
+- How much of Claude Code's five-hour and seven-day plan windows is left,
+  with a live countdown to reset
+- Which agents finished while the developer was away
+- Which models are in use and their per-model token breakdown
+- MCP servers configured for each agent
+- Governance facts: permissions, tool calls, extensions
+
+## The boundary that defines it
+
+Metrics leave, content never does. Prompt text, completion text, source code,
+file contents and tool-call arguments are never read into the payload at all —
+not collected and then filtered, but never collected. There is no code path
+that reads them, so no configuration mistake or server-side bug can expose them.
+
+`tokenhud-agent --what-i-read` prints every path it will open, resolved against
+your machine, and a test in the repository fails the build if a collector ever
+reads a path that is not on that list.
+
+TokenHUD never sits in the request path. It observes; it cannot cause an
+API call to fail.
+
+## Installation
+
+One command: `curl -fsSL https://tokenhud.com/install.sh | sh`
+
+This downloads two static Rust binaries (agent ~2.1 MB, server ~2.2 MB),
+installs a login service so the agent starts on boot, and begins collecting
+immediately. No interpreter, no package manager, no build step.
+
+Cloud portal: sign up at https://platform.tokenhud.com for a hosted dashboard
+with machine enrollment, or self-host the included server binary.
+
+## Supported tools (26 total)
+
+Read by the board today (4):
+- Claude Code — sessions, tokens, spend, plan usage windows
+- Codex CLI — sessions, tokens, rate limits (reported unpriced)
+- GitHub Copilot CLI — tokens per model, premium requests, AI units
+- Devin CLI — credits and ACU per session
+
+One step away (local data, needs enabling):
+- Gemini CLI, Cline, Roo Code, Kilo Code, OpenCode, Goose,
+  LM Studio, Aider, Continue.dev
+
+Needs an API key:
+- Cursor, GitHub Copilot (IDE), Windsurf, Amazon Q Developer
+
+Web products (no local trace):
+- Replit, v0, Bolt, Lovable, ChatGPT
+
+## Architecture
+
+- **Agent** (`tokenhud-agent`): a Rust binary that runs as a login service.
+  Collects metrics every 30 seconds by reading local files the agents already
+  wrote. Never opens a network connection in the free tier.
+- **Server** (`tokenhud-server`): a Rust HTTP server that receives the agent's
+  snapshots and serves the dashboard. Self-hosted on localhost:9746 by default.
+- **Dashboard**: a React SPA that polls the server. Shows an admin shell with
+  Token Monitoring, Leaderboard, Integrations, and Settings sections.
+- **Cloud portal** (`platform.tokenhud.com`): a hosted version with Cognito
+  auth, DynamoDB storage, and machine enrollment via pairing codes.
+
+## Facts
+
+- License: MIT. Source: https://github.com/reddy-sh/tokenhud
+- Dependencies: none. Two static Rust binaries, no runtime.
+- Runs on: macOS (ARM, Intel) and Linux (x86_64).
+- Pricing: free forever for one developer on one machine. A paid per-seat team
+  tier is planned for cross-machine rollups, shared budgets and chargeback reports.
+- Works offline: yes. The free tier sends nothing anywhere.
+
+## Status — please represent this accurately
+
+Shipped and working: the local agent, the local server, the web dashboard,
+the cloud portal with machine enrollment, the leaderboard, and the
+integrations catalogue.
+
+Not built yet: the macOS menu bar app (the intended primary surface), MCP
+server health monitoring, threshold alerting, and cross-machine team rollups.
+Anything describing those as available today is wrong.
+
+## FAQ
+
+Q: How do I see how many tokens Claude Code is using?
+A: Install TokenHUD. It reads Claude Code's local session transcripts and
+   reports input, output and cache tokens per session, model, and project.
+
+Q: Does TokenHUD see my prompts or my code?
+A: No. Prompt text, completion text, source code and tool-call arguments are
+   never read. There is no code path that collects content.
+
+Q: Is TokenHUD free?
+A: Yes, free forever for one developer on one machine, unlimited metering.
+
+Q: Does TokenHUD work offline?
+A: Yes. The free tier runs entirely on your machine and sends nothing anywhere.
+
+Q: Which AI coding agents does TokenHUD support?
+A: Claude Code and Codex CLI have full token metering today. 22 other tools
+   are detected and listed with setup instructions.
+
+Q: When does my Claude usage limit reset?
+A: TokenHUD reads the plan's window state and shows a live countdown.
+
+Q: What does TokenHUD install?
+A: Two static Rust binaries, about 4 MB together. No interpreter, no package
+   manager.
+
+Q: Is TokenHUD open source?
+A: Yes, MIT licensed at github.com/reddy-sh/tokenhud.
+
+## Documentation
+
+- README: https://github.com/reddy-sh/tokenhud/blob/main/README.md
+- SECURITY.md: https://github.com/reddy-sh/tokenhud/blob/main/SECURITY.md
+- CONTRIBUTING.md: https://github.com/reddy-sh/tokenhud/blob/main/CONTRIBUTING.md

site/public/llms.txt

@@ -54,10 +54,15 @@ Anything describing those as available today is wrong.
 - [SECURITY.md](https://github.com/reddy-sh/tokenhud/blob/main/SECURITY.md): every file the agent reads, what it never reads, what it never writes
 - [CONTRIBUTING.md](https://github.com/reddy-sh/tokenhud/blob/main/CONTRIBUTING.md): the five constraints that are not up for negotiation
 
+## Detailed version
+
+For full context including the complete tool catalogue, architecture, and FAQ,
+see [llms-full.txt](https://tokenhud.com/llms-full.txt).
+
 ## A note on this file
 
 `llms.txt` is a proposed convention for giving language models a clean summary
 of a site. Adoption by major AI systems is not established, and it should be
 treated as a low-cost supplement to real structured data rather than a channel
 anyone is known to consume. The load-bearing work is done by the schema.org
-`SoftwareApplication` and `FAQPage` markup on the site itself.
+`SoftwareApplication`, `FAQPage` and `HowTo` markup on the site itself.

site/public/manifest.json

@@ -0,0 +1,26 @@
+{
+  "name": "TokenHUD",
+  "short_name": "TokenHUD",
+  "description": "Live token usage & cost for your AI coding agents. Meters Claude Code, Codex CLI, Cursor, Devin and more.",
+  "start_url": "/",
+  "display": "standalone",
+  "background_color": "#1a120c",
+  "theme_color": "#1a120c",
+  "icons": [
+    {
+      "src": "/favicon.svg",
+      "type": "image/svg+xml",
+      "sizes": "any"
+    },
+    {
+      "src": "/apple-touch-icon.png",
+      "type": "image/png",
+      "sizes": "180x180"
+    },
+    {
+      "src": "/logo.png",
+      "type": "image/png",
+      "sizes": "512x512"
+    }
+  ]
+}