chore: modernize README — clean, professional, matches website

Author: Vitalcheffe

README.md

@@ -6,95 +6,85 @@ Local-first. Zero-trust. Auditable by design.
 
 ---
 
-## The problem nobody talks about
+## The problem
 
-Every AI coding agent on the market does the same thing: it takes your code, sends it to a cloud API, and hopes for the best. No sandbox. No audit trail. No cryptographic guarantees. You are trusting an external server with your proprietary source code, your API keys, your infrastructure.
+Every AI coding agent sends your code to the cloud. No sandbox. No audit trail. No cryptographic guarantees. You are trusting strangers with your proprietary code.
 
-Kilo Code raised $8 million and has 1.5 million users. Cline has 5 million installations. Cursor raised at a $9 billion valuation. They are all built on the same foundation: send your code to the cloud and pray.
+Kilo Code raised $8M. Cline has 5M installs. Cursor is worth $9B. They all do the same thing: send your code to a server you don't control.
 
-WRAP NEBULA is built on the opposite foundation: **never trust, always verify**.
+WRAP NEBULA does the opposite.
 
-## What makes this different
+## What makes it different
 
 | | Kilo | Cline | Cursor | **WRAP** |
 |---|---|---|---|---|
-| Code sandbox | no | no | no | **V8 isolate** |
+| Sandbox | no | no | no | **V8 isolate** |
 | Audit trail | no | no | no | **Ed25519** |
-| PII protection | no | no | no | **Auto-redact** |
+| PII redaction | no | no | no | **Automatic** |
 | Local-first | partial | partial | no | **Default** |
-| Zero API keys | no | no | no | **Ollama** |
+| Free forever | no | no | no | **Yes** |
 | Telegram | no | no | no | **Built-in** |
-| Open source | yes | yes | no | **MIT** |
-
-## How it works
-
-Every message flows through a pipeline where compromising one layer does not compromise the others:
-
-1. **Input** - Telegram, CLI, VS Code, or web
-2. **Input Sanitizer** - blocks prompt injection (10 patterns), SQL injection (6), XSS (5). PII redacted automatically
-3. **Rust Governor** - separate process. Binary-level audit signing. Policy enforcement
-4. **SOUL.md** - agent personality, skills, and rules in simple markdown
-5. **LLM** - Ollama (local, free), Claude, or GPT-4. Auto-detected
-6. **Skills** - 14 sandboxed executors
-7. **Memory** - SQLite, local only, never transmitted
-8. **Response** - signed, audited, delivered
-
-## The 14 skills
-
-- web.search - DuckDuckGo scraping, no API key
-- files.read/write/list - path-restricted filesystem ops
-- code.run - V8 sandboxed execution
-- code.edit - diff-based editing with backup
-- code.search - grep, find, symbol search
-- terminal.run - shell commands with safety checks
-- reminder.set/list - natural language dates, SQLite storage
-- git.status - structured git output
-- calendar.read - local .ics reader
-- email.summary - local .mbox reader
-- project.context - smart file inclusion for LLM
-
-Every skill is hash-verified before loading. Sandbox blocks dangerous patterns.
 
 ## Install
 
 ```
+# Authenticate
+nebula auth login anthropic
+
+# Install
 curl -fsSL https://raw.githubusercontent.com/Vitalcheffe/Wrap/main/install.sh | bash
-nebula init
-nebula start
+
+# Start (Ollama must be running)
+ollama serve && ollama pull llama3 && nebula start
 ```
 
-Requires Ollama running locally. No API keys needed.
+## How it works
 
-## CLI
+Every message flows through a pipeline where compromising one layer does not compromise the others:
 
-- nebula init - setup wizard
-- nebula start - start the agent
-- nebula stop - stop the agent
-- nebula status - show config
-- nebula doctor - health check
-- nebula skill - manage skills
-- nebula agents - multi-agent info
+1. Input — Telegram, CLI, VS Code
+2. Sanitizer — blocks injection, redacts PII
+3. Rust Governor — separate process, policy enforcement
+4. SOUL.md — agent personality in markdown
+5. LLM — Ollama / Claude / GPT-4
+6. Skills — 14 sandboxed executors
+7. Memory — SQLite, local only
+8. Response — signed, audited
+
+## Skills
+
+- web.search — DuckDuckGo scraping, no API key
+- files.read/write/list — path-restricted filesystem
+- code.run — V8 sandboxed execution
+- code.edit — diff-based editing with backup
+- code.search — grep, find, symbol search
+- terminal.run — shell with safety checks
+- reminder.set/list — natural language dates
+- git.status — structured git output
+- calendar.read — local .ics reader
+- email.summary — local .mbox reader
+- project.context — smart file inclusion
 
-## VS Code Extension
+## CLI
 
-- Chat panel - talk to your agent in VS Code
-- Right-click actions - select code, explain/fix/review
-- Audit trail tree - every signed action in real-time
-- Status bar - agent online/offline
+- nebula init — setup wizard
+- nebula start — start the agent
+- nebula stop — stop the agent
+- nebula status — show status
+- nebula doctor — health check
+- nebula auth login — authenticate
+- nebula auth list — show credentials
+- nebula skill create — create a skill
 
 ## Security
 
-### Input Sanitizer
-10 prompt injection patterns, 6 SQL patterns, 5 XSS patterns. PII auto-redacted: SSN, credit cards, emails, phones.
+Input Sanitizer: 10 prompt injection, 6 SQL, 5 XSS patterns. PII redacted.
 
-### V8 Sandbox
-Temp HOME, minimal PATH, no network, 128MB memory, 30s timeout. Dangerous patterns blocked.
+V8 Sandbox: temp dir, no network, 128MB, 30s timeout.
 
-### Ed25519 Audit Trail
-Every action signed. Hash-chained. Verify all with one function call. Immutable.
+Ed25519 Audit Trail: every action signed. Hash-chained. Verify all.
 
-### SQLite Memory
-Local only. Never transmitted. Never synced.
+SQLite Memory: local only. Never transmitted.
 
 ## Testing
 
@@ -105,19 +95,10 @@ npx vitest run    # 46/46 passing
 npx tsc --noEmit  # 0 errors
 ```
 
-## Roadmap
-
-- v8 Core agent, sandbox, audit, Telegram, CLI - Shipped
-- v9 VS Code extension - Shipped
-- v10 Plugin system - Shipped
-- v11 Multi-agent - Shipped
-- v12 Published extension, community - Active
-- v13 Enterprise, SSO, compliance - Planned
-
 ## License
 
 MIT
 
 ---
 
-Built by VitalCheffe - 16, Casablanca, Morocco
+Built by VitalCheffe — 16, Casablanca, Morocco