Lead the docs with what an agent gets, in the words buyers search

README.md

@@ -1,6 +1,6 @@
 # Proofable Docs
 
-Give AI agents identity, scoped access, and proof of what they did. Add verification and reusable proof to your app.
+Put AI agents on real work without handing over your keys. These are the docs for Proofable: agent permissions, approvals, signed proof, and human and domain checks you can reuse in your app.
 
 Published site: [docs.proofable.me](https://docs.proofable.me)
 

agents/agent-delegation.mdx

@@ -1,7 +1,7 @@
 ---
-title: Agent authority
-sidebarTitle: Authority
-description: "Grant an agent specific allowed actions, a per-payment spend cap, and an expiry date. Revoke or narrow its authority at any time."
+title: Agent permissions
+sidebarTitle: Permissions
+description: "Grant an AI agent specific allowed actions, a per-payment spend cap and an expiry date. Narrow or revoke it at any time."
 icon: handshake
 ---
 

agents/index.mdx

@@ -1,7 +1,7 @@
 ---
 title: "Agents"
 sidebarTitle: "Agents"
-description: "Give each agent its own identity and limits. The controls stay with the agent across models, tools, runtimes, and MCP."
+description: "Give each AI agent its own identity and a permission: the tools it can use, what it can spend, and when access ends. It goes with the agent into every client."
 icon: robot
 "og:image": "/images/og-default.png"
 "twitter:image": "/images/og-default.png"

connections.mdx

@@ -1,7 +1,7 @@
 ---
 title: "Connect"
 sidebarTitle: "Connections"
-description: "Your context and control, anywhere."
+description: "Link an app once. Your agents use it under their permission and never see the credentials."
 icon: plug
 ---
 

docs.json

@@ -59,10 +59,11 @@
     }
   },
   "navbar": {
+    "links": [{ "label": "Sign in", "href": "https://proofable.me/verify?intent=login" }],
     "primary": {
       "type": "button",
-      "label": "Sign in",
-      "href": "https://proofable.me/verify?intent=login"
+      "label": "Start free",
+      "href": "https://proofable.me/connect"
     }
   },
   "navigation": {

evidence.mdx

@@ -1,5 +1,5 @@
 ---
-title: "Evidence"
+title: "Agent security benchmarks"
 sidebarTitle: "Evidence"
 description: "How much agent time and spend Proofable saves, and what it stops, measured against the same work done without it."
 icon: certificate

index.mdx

@@ -1,24 +1,26 @@
 ---
-title: "Get Started"
+title: "Get started with Proofable"
 sidebarTitle: "Get Started"
-description: "Set limits on what AI agents can do, verify people and accounts, and keep proof of what happened."
-"og:title": "Grant access with proof your app can check again"
-"twitter:title": "Grant access with proof your app can check again"
+description: "Connect Proofable to Claude, Cursor, Codex or VS Code with one command, give an agent a permission, and check its signed proof."
+"og:title": "Put AI agents on real work without handing over your keys"
+"twitter:title": "Put AI agents on real work without handing over your keys"
 icon: book
 "og:image": "/images/og-default.png"
 "twitter:image": "/images/og-default.png"
 ---
 
-Proofable lets people, businesses, and AI agents prove something once and reuse the proof anywhere. Each agent gets a scoped permission instead of your keys, every action is checked before it runs, and every run leaves a signed proof anyone can verify offline.
+Proofable puts AI agents on real work without your keys. Each agent gets a permission for its job: the tools it can use, what it can spend on each payment, and when its access ends. Every action is checked before it runs, risky steps can wait for your approval, and every run leaves a signed proof anyone can verify offline.
 
-Add it to your AI tool, or build it into your product.
+People can also prove once that they are human or own a domain, and reuse that proof at every later check.
+
+Start in your AI tool, or build it into your product. Free with 500 credits a month.
 
 <CardGroup cols={3}>
   <Card title="Install Proofable" icon="plug" iconType="solid" href="/mcp/setup">
-    Add Proofable to any MCP client.
+    One connection for Claude, Cursor, Codex and VS Code.
   </Card>
-  <Card title="Protect product access" icon="shield-check" iconType="solid" href="/integrations/server">
-    Check the proof before your server grants access.
+  <Card title="Give an agent a permission" icon="shield-check" iconType="solid" href="/agents/agent-delegation">
+    Its tools, its spend cap, its end date. Nothing else runs.
   </Card>
   <Card title="Verify a person or account" icon="key" iconType="solid" href="/verification/hosted">
     Run a hosted check and keep the proof.

mcp/setup.mdx

@@ -1,7 +1,7 @@
 ---
-title: "Setup"
+title: "Set up Proofable in Claude, Cursor, Codex or VS Code"
 sidebarTitle: "Setup"
-description: "Connect Proofable to an MCP client, sign in, and start with your current profile and proofs."
+description: "Add the Proofable MCP server to your client with one command, sign in, and start with your profile and the proofs you can reuse."
 icon: plug
 ---
 

package.json

@@ -1,7 +1,7 @@
 {
   "name": "@proofable/docs",
   "private": true,
-  "description": "Docs for verification, gates, proofs, agents, and MCP.",
+  "description": "Docs for Proofable: AI agent permissions, approvals, signed proof, verification and MCP.",
   "license": "Apache-2.0",
   "type": "module",
   "engines": {

security.mdx

@@ -1,5 +1,6 @@
 ---
-title: "Security"
+title: "AI agent security"
+sidebarTitle: "Security"
 description: "AI agent security and governance. An agent cannot act beyond what you allowed, and a proof never exposes the data behind it."
 icon: lock
 ---

use-cases/prove-what-happened.mdx

@@ -1,7 +1,7 @@
 ---
-title: "Prove what happened"
+title: "AI agent audit trail"
 sidebarTitle: "Prove what happened"
-description: "Keep a record of who acted and what was allowed."
+description: "Every permission, check and agent action becomes a signed proof of who acted, what was allowed and what happened, without exposing the data behind it."
 icon: receipt
 ---