cbrowser 19.6.0 → 20.0.0

package/README.md

@@ -1,62 +1,78 @@
-# CBrowser — Cognitive Browser Automation
+# CBrowser: Cognitive Browser Automation
 
-> **The browser automation that thinks like your users.** 122 MCP tools. 25 cognitive traits. 11 disability personas. The only framework that predicts when users give up.
+> **Browser automation that thinks like your users.** 130 MCP tools, 21 built-in personas, a CLI, and a library. It predicts where a real person gives up before they do.
 
 [![npm version](https://img.shields.io/npm/v/cbrowser.svg)](https://www.npmjs.com/package/cbrowser)
 [![Documentation](https://img.shields.io/badge/Docs-cbrowser.ai-blue.svg)](https://cbrowser.ai/docs)
 [![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)
-[![MCP Ready](https://img.shields.io/badge/MCP-122%20Tools-blue)](https://modelcontextprotocol.io)
-[![TypeScript](https://img.shields.io/badge/TypeScript-5.3-blue.svg)](https://www.typescriptlang.org/)
-[![Node](https://img.shields.io/badge/Node-%3E%3D18-green.svg)](https://nodejs.org/)
+[![MCP Ready](https://img.shields.io/badge/MCP-130%20Tools-blue)](https://modelcontextprotocol.io)
+[![TypeScript](https://img.shields.io/badge/TypeScript-strict-blue.svg)](https://www.typescriptlang.org/)
+[![Node](https://img.shields.io/badge/Node-%3E%3D20.9-green.svg)](https://nodejs.org/)
 
 ```bash
-npm install cbrowser
-npx cbrowser cognitive-effort --url "https://your-site.com" --persona first-timer
+npx cbrowser agent-ready-audit "https://your-site.com" --html
 ```
 
 ---
 
-## 5-Minute Quickstart
-
-Get from zero to your first cognitive insight in 3 commands:
+## Quickstart
 
 ```bash
-# 1. Install CBrowser
-npm install cbrowser
+# 1. Install (Node.js 20.9 or newer). Installing also downloads Chromium.
+npm install -g cbrowser
+
+# 2. Check the environment
+cbrowser doctor
 
-# 2. Install the browser engine (~150MB)
-npx playwright install chromium
+# 3. Free, keyless audits
+cbrowser agent-ready-audit "https://your-site.com" --html
+cbrowser empathy-audit "https://your-site.com" --personas cognitive-adhd,low-vision-magnified --html
 
-# 3. Run your first cognitive audit
-npx cbrowser cognitive-effort --url "https://your-site.com" --persona first-timer
+# 4. A simulated visitor (needs an Anthropic API key)
+cbrowser config set-api-key <your-anthropic-key>
+cbrowser cognitive-journey --persona first-timer --start "https://your-site.com" --goal "sign up"
 ```
 
-**What you get:** A cognitive transport score (0-1), abandonment risk percentage, and the UX bottleneck — in under 30 seconds.
+If the browser download was skipped (for example with `--ignore-scripts`), run `npx playwright install chromium`.
 
-**No API key needed** for basic commands (navigate, screenshot, click, extract, explore). Cognitive journeys require an Anthropic API key:
+**Stuck? Ask the docs.** `cbrowser ask` sends your question to cbrowser.ai and prints an AI answer grounded in the documentation, with its sources. No account or API key is needed; it is rate-limited.
 
 ```bash
-npx cbrowser config set-api-key <your-anthropic-key>
+cbrowser ask "Which persona should I use for a screen reader user?"
 ```
 
-**Check your environment:** Run `npx cbrowser doctor` to verify everything is set up correctly.
+The `question_answer` MCP tool does the same from Claude, including on the local `npx cbrowser mcp-server`.
 
-**Accessibility:** CBrowser supports `--no-color` (or `NO_COLOR` env var), `--plain` (no emoji/decorations), and `--json-output` (structured JSON) for screen readers, CI pipelines, and scripting.
+### What needs an API key
+
+| Free, no key | Needs an Anthropic API key (each step is a model call) |
+|---|---|
+| `navigate`, `click`, `fill`, `type`, `scroll`, `screenshot`, `extract`, `evaluate` | `cognitive-journey` |
+| `agent-ready-audit`, `ai-benchmark`, `empathy-audit` (without `--goal`) | `explore` (a short cognitive journey) |
+| `test-suite`, `repair-tests`, `flaky-check`, `hunt-bugs`, `chaos` | `compare-personas` (one journey per persona) |
+| `ai-visual`, `cross-browser`, `responsive`, `ab`, `perf`, `a11y audit` | `competitive-benchmark` (one journey per site) |
+| `capture`, `har`, `network`, `console`, `daemon`, `ask` | `empathy-audit --goal` (the journey part) |
+
+Since 20.0 nothing falls back to a keyless guess. The old keyless modes of `explore` and `competitive-benchmark` reported goals as reached when a goal word, "the" included, appeared on the page, so they now refuse without a key. A journey that is replayed from a saved trace is free (see [Replay](#replay-a-journey-for-free)).
+
+**Output for scripts and screen readers:** `--no-color` (or `NO_COLOR`), `--plain` (no emoji or box drawing; pass and fail marks become `[PASS]` and `[FAIL]`), and `--json-output` (one JSON envelope on stdout).
 
 ---
 
 ## Table of Contents
 
-- [5-Minute Quickstart](#5-minute-quickstart)
+- [Quickstart](#quickstart)
+- [Upgrading to 20.0.0](#upgrading-to-2000)
 - [Cognitive Transport Chain](#cognitive-transport-chain)
-- [Visual Overlays & Attention Analysis](#visual-overlays--attention-analysis)
+- [Visual Overlays and Attention Analysis](#visual-overlays-and-attention-analysis)
 - [AI Friendliness Suite](#ai-friendliness-suite)
 - [Cognitive User Simulation](#cognitive-user-simulation)
 - [Accessibility Empathy Testing](#accessibility-empathy-testing)
 - [Constitutional AI Safety](#constitutional-ai-safety)
-- [Site Knowledge System](#site-knowledge-system)
+- [Browser Control from the CLI](#browser-control-from-the-cli)
 - [Natural Language Testing](#natural-language-testing)
 - [Visual Testing](#visual-testing)
+- [Screen Capture](#screen-capture)
 - [Competitive UX Benchmark](#competitive-ux-benchmark)
 - [Geo Proxy](#geo-proxy)
 - [MCP Server Integration](#mcp-server-integration)
@@ -64,14 +80,47 @@ npx cbrowser config set-api-key <your-anthropic-key>
 - [API Reference](#api-reference)
 - [Configuration](#configuration)
 - [Enterprise Edition](#enterprise-edition)
-- [Contributing](#contributing)
 - [License](#license)
 
 ---
 
+## Upgrading to 20.0.0
+
+20.0 is a breaking release. Most of it makes commands that used to succeed silently either do what they claim or fail out loud. The full list is in [CHANGELOG.md](CHANGELOG.md).
+
+**Runtime**
+- Node.js 20.9 or newer. Node 18 is end of life and sharp 0.35 dropped it; stay on cbrowser 19.x if you need Node 18.
+
+**Behaviour you may notice**
+- The default page size is **1920x1080** (was 1280x800), the size `desktop` means everywhere else. Set `CBROWSER_VIEWPORT_WIDTH`/`CBROWSER_VIEWPORT_HEIGHT` or `viewport` in the config file to keep the old size.
+- Pages render in the host's locale and timezone unless you set `--locale`/`--timezone`, `CBROWSER_LOCALE`/`CBROWSER_TIMEZONE` or the config file (it was en-US and America/Denver). Env vars and the config file are no longer overridden by built-in values.
+- A one-off `--device` or `--geo` no longer sticks to later commands.
+- `explore`, `CBrowser.journey()` and `competitive-benchmark` need an Anthropic API key. Their keyless heuristic modes are gone (see above).
+- Commands that fail exit 1. That includes unknown subcommands, failed clicks and navigations through the daemon, `record start/stop/save` (which never recorded across commands and now refuse), `perf audit` and `visual compare` failures, and `daemon status` when nothing is running.
+- Commands that check their flags reject flags they never read. `--quiet` is gone, since nothing read it.
+- `--json-output`: `extract`, `analyze` and `network list` put their data in `data`. Other commands still return captured text lines in `data.output`.
+- `--plain` and `--json-output` text keep pass, fail and warning marks as `[PASS]`, `[FAIL]` and `[WARN]`.
+- With `--html` and an `--output` that does not end in `.json`, the HTML report goes to `<name>.report.html` instead of overwriting the JSON.
+- `repair-tests` saves its report with `--json-out <file>`; `--json` prints it.
+- Size names: `mobile` is 393x852, `tablet` 834x1194 and `desktop` 1920x1080 in every tool (the CLI's `responsive` and MCP `responsive_test` used smaller sizes). The old sizes remain as `tablet-sm` (768x1024) and `desktop-md` (1440x900); `desktop-xl` is 2560x1440.
+
+**MCP**
+- `npx cbrowser mcp-server` (stdio) now registers the same tool set as the hosted server, 130 tools plus 7 local-only ones. It had its own copy of 82 tools.
+- Red-zone refusals from `click`, `smart_click`, `press_key`, `type_text`, `drag` and `evaluate_script` are tool errors with one shape (`isError`, `zone: "red"`, `howToRun`). On the hosted server a refused call is refunded.
+- `get_network_requests` returns the newest 50. `manage_cookies` `set` writes a session cookie unless `expires` is given.
+- `compare_personas_record_result` and `compare_personas_summarize` are removed; nothing could create the session they needed.
+- The five Enterprise marketing stubs use the Enterprise server's tool names and parameters.
+
+**CI**
+- The GitHub Action's `sensitivity` input is removed (test-suite never read it).
+
+Upgrading from 18.x? The 19.0 change was the agent-ready `accessibility` axis becoming `agentPerceivability`: `score.accessibility` is `score.agentPerceivability`, `summary.dynamicContentCount` is `summary.deferredLoadingPatterns`, and `comparison.bestAccessibility` is `comparison.bestAgentPerceivability`.
+
+---
+
 ## Cognitive Transport Chain
 
-The eight-layer Sequential Transport Chain computes exactly how hard your page is for any persona - and predicts whether they'll abandon. Each layer answers one question about the visitor:
+The eight-layer Sequential Transport Chain computes how hard a page is for a given persona, and predicts whether they abandon. Each layer answers one question about the visitor:
 
 | Layer | The Question It Answers | Research Basis |
 |-------|------------------------|----------------|
@@ -84,66 +133,60 @@ The eight-layer Sequential Transport Chain computes exactly how hard your page i
 | **Readability** | Can they decode the words on the page? | Perry & Zorzi (2013) reading model |
 | **Reading Attention** | Can they hold the line long enough to finish? | Sustained attention (split from Readability) |
 
-The chain is *sequential* because attention is a budget, not a checklist. Each layer's cost is paid from whatever the layers before it left behind: a visitor who burned working memory just finding the button has less left for the form, and a persona who arrives at your pricing table already frustrated reads it differently than one who got there clean. That's why a page that exhausts people early scores differently from one that is uniformly hard, and why the bottleneck layer matters more than the average.
+The chain is *sequential* because attention is a budget, not a checklist. Each layer's cost is paid from whatever the layers before it left behind: a visitor who burned working memory just finding the button has less left for the form, and a persona who arrives at your pricing table already frustrated reads it differently from one who got there clean. That is why a page that exhausts people early scores differently from one that is uniformly hard, and why the bottleneck layer matters more than the average.
 
-Two layers were split out of their neighbors because they fail for different reasons. **Motor Procedure** is separate from **Motor** because hitting a target and completing a sequence are different failures - you can land every click and still lose your place halfway through a form. **Reading Attention** is separate from **Readability** because decoding text and holding your place in it are different abilities - an ADHD reader typically decodes fine and loses the thread, and a single "reading" score can't say that.
+Two layers were split out of their neighbours because they fail for different reasons. **Motor Procedure** is separate from **Motor** because hitting a target and completing a sequence are different failures: you can land every click and still lose your place halfway through a form. **Reading Attention** is separate from **Readability** because decoding text and holding your place in it are different abilities: an ADHD reader typically decodes fine and loses the thread, and a single "reading" score cannot say that.
 
-```bash
-npx cbrowser cognitive-effort --url "https://your-site.com" --persona first-timer
-# → CTC: 0.76, abandonment risk: 42%, bottleneck: decision
-```
+The chain runs through the MCP tool `cognitive_effort` (there is no CLI command for it):
 
-Returns: total CTC score, per-layer breakdown, bottleneck identification, abandonment risk percentage, and motor accessibility overlay.
+```text
+cognitive_effort(url: "https://your-site.com", persona: "first-timer")
+-> CTC score, per-layer breakdown, bottleneck layer, abandonment risk, motor overlay
+```
 
 ---
 
-## Visual Overlays & Attention Analysis
-
-See where users look, what they miss, and whether your CTAs capture attention — per persona:
+## Visual Overlays and Attention Analysis
 
-```bash
-# Full visual cognitive story — 4 annotated images + narrative
-npx cbrowser visual-cognitive-story --url "https://your-site.com" --persona cognitive-adhd
+See where users look, what they miss, and whether your calls to action capture attention, per persona. These are MCP tools:
 
-# Attention heatmap — where does this persona look?
-npx cbrowser attention-analysis --url "https://your-site.com" --persona first-timer
-
-# Compare how two personas experience the same page
-npx cbrowser attention-compare --url "https://your-site.com" --persona-a first-timer --persona-b power-user
-```
+| MCP tool | What it returns |
+|---|---|
+| `visual_cognitive_story` | Four annotated images and a narrative for one persona |
+| `attention_analysis` | Predicted attention heatmap for one persona |
+| `attention_compare` | How two personas experience the same page |
+| `journey_heatmap_gif` | An animated heatmap of a journey |
 
-**Overlays generated:**
-- **Attention heatmap** — CIE-Lab saliency on grayscale base (red = high attention)
-- **Motor accessibility** — Green/yellow/red per element by P(hit)
-- **Attention quality** — CTAs green, distractors red, headings blue
-- **Comparison diff** — Blue = persona A attention, red = persona B
+**Overlays:** attention heatmap (CIE-Lab saliency on a grayscale base, red is high attention), motor accessibility (green, yellow or red per element by probability of a hit), attention quality (calls to action green, distractors red, headings blue) and a comparison diff (blue is persona A, red is persona B).
 
-**Metrics:** CTA capture rate, value prop salience, distractor ratio, quality score (0-100), entropy.
+**Metrics:** CTA capture rate, value-prop salience, distractor ratio, quality score (0-100), entropy.
 
 ---
 
 ## AI Friendliness Suite
 
-Five tools to make your site ready for the AI agent era:
+Make your site work for AI agents as well as people:
 
 | Tool | What It Does |
 |------|-------------|
-| `agent_ready_audit` | Score site on findability, stability, agent-perceivability, semantics (A-F grade) |
-| `ai_benchmark` | Compare AI-friendliness across competitor URLs |
-| `webmcp_ready_audit` | Audit MCP server for WebMCP compatibility |
-| `remediation_patches` | Generate actionable code fixes for audit findings |
-| `llms_txt_generate` | Generate AI-readable llms.txt site description |
+| `agent_ready_audit` | Scores findability, stability, agent-perceivability and semantics (A-F grade) |
+| `ai_benchmark` | Compares AI-friendliness across competitor URLs |
+| `webmcp_ready_audit` | Audits a site for WebMCP readiness |
+| `remediation_patches` | Generates code fixes for audit findings |
+| `llms_txt_generate` | Generates an AI-readable llms.txt for a site |
 
 ```bash
-npx cbrowser agent-ready-audit "https://your-site.com" --html
-npx cbrowser ai-benchmark "https://site-a.com,https://site-b.com"
+cbrowser agent-ready-audit "https://your-site.com" --html
+cbrowser ai-benchmark --urls "https://site-a.com,https://site-b.com" --html
 ```
 
+The agent-perceivability axis measures whether a machine can perceive your elements (ARIA present, elements carry text). It is not a WCAG conformance score; run `empathy-audit` for that, and expect the two to differ.
+
 ---
 
 ## Cognitive User Simulation
 
-**26 research-backed cognitive traits + 10 motivational values** model how real users think, struggle, and give up:
+Research-backed cognitive traits and motivational values model how real users think, struggle and give up:
 
 | Trait | Research Basis | What It Models |
 |-------|---------------|----------------|
@@ -151,238 +194,235 @@ npx cbrowser ai-benchmark "https://site-a.com,https://site-b.com"
 | **Working Memory** | Miller (1956) | 7±2 item capacity; affects form complexity tolerance |
 | **Comprehension** | Card, Moran & Newell (1983) | UI convention understanding; GOMS model timing |
 | **Risk Tolerance** | Kahneman & Tversky (1979) | Prospect theory; loss aversion affects CTA clicks |
-| **Self-Efficacy** | Bandura (1977) | Belief in ability to solve problems; low = faster abandonment |
-| **Satisficing** | Simon (1956) | Accept "good enough" vs. optimize; 50% faster decisions |
-| **Trust Calibration** | Fogg (2003) | 8 trust signals; affects click-through by 40% |
-| **Information Foraging** | Pirolli & Card (1999) | "Scent" following behavior; predicts navigation patterns |
-| **Social Proof** | Cialdini (2001) | Influence of reviews, ratings, popularity indicators |
+| **Self-Efficacy** | Bandura (1977) | Belief in ability to solve problems; low means faster abandonment |
+| **Satisficing** | Simon (1956) | Accept "good enough" or optimize; faster decisions |
+| **Trust Calibration** | Fogg (2003) | Trust signals; affects click-through |
+| **Information Foraging** | Pirolli & Card (1999) | "Scent" following; predicts navigation patterns |
+| **Social Proof** | Cialdini (2001) | Influence of reviews, ratings and popularity indicators |
 | **FOMO** | Przybylski et al. (2013) | Fear of missing out; urgency and scarcity responses |
 
-*See [Trait Index](https://cbrowser.ai/docs/Trait-Index) for all 25 traits including: Persistence, Resilience, Curiosity, Change Blindness, Anchoring Bias, Time Horizon, Attribution Style, Metacognitive Planning, Procedural Fluency, Transfer Learning, Authority Sensitivity, Emotional Contagion, Mental Model Rigidity, Interrupt Recovery, Reading Tendency, and Site Familiarity.*
+The [Trait Index](https://cbrowser.ai/docs/Trait-Index) lists every trait with its citations, including Persistence, Resilience, Curiosity, Change Blindness, Anchoring Bias, Time Horizon, Attribution Style, Metacognitive Planning, Procedural Fluency, Transfer Learning, Authority Sensitivity, Emotional Contagion, Mental Model Rigidity, Interrupt Recovery, Reading Tendency and Site Familiarity.
 
 **Full documentation:** [Research Methodology](https://cbrowser.ai/docs/Research-Methodology) · [Trait Index](https://cbrowser.ai/docs/Trait-Index) · [Bibliography](https://cbrowser.ai/docs/Bibliography)
 
-### 17 Built-in Personas
+### 21 Built-in Personas
 
-**6 general:** first-timer, power-user, mobile-user, screen-reader-user, elderly-user, impatient-user
+**10 general:** power-user, first-timer, mobile-user, screen-reader-user, elderly-user, impatient-user, distracted-user, careful-reader, explorer, task-focused
 
 **11 accessibility:** motor-impairment-tremor, low-vision-magnified, cognitive-adhd, dyslexic-user, deaf-user, elderly-low-vision, color-blind-deuteranopia, autism-spectrum, intellectual-disability, aphasia-receptive, dyscalculia
 
+`cbrowser persona list` shows them with any custom personas you have made.
+
 ### Values System
 
-Beyond cognitive traits, CBrowser models **10 motivational values** from three psychological frameworks:
+Beyond cognitive traits, CBrowser models motivational values from three psychological frameworks:
 
 | Framework | Research Basis |
 |-----------|---------------|
-| **Schwartz's Universal Values** | Schwartz (1992) — 10 core values: Power, Achievement, Hedonism, Stimulation, Self-Direction, Universalism, Benevolence, Tradition, Conformity, Security |
-| **Self-Determination Theory** | Deci & Ryan (1985) — Autonomy, Competence, Relatedness |
-| **Maslow's Hierarchy** | Maslow (1943) — 5 need levels |
+| **Schwartz's Universal Values** | Schwartz (1992): Power, Achievement, Hedonism, Stimulation, Self-Direction, Universalism, Benevolence, Tradition, Conformity, Security |
+| **Self-Determination Theory** | Deci & Ryan (1985): Autonomy, Competence, Relatedness |
+| **Maslow's Hierarchy** | Maslow (1943): 5 need levels |
 
-See [Values Framework](https://cbrowser.ai/docs/Values-Framework) for full documentation.
+See [Values Framework](https://cbrowser.ai/docs/Values-Framework).
 
-### Custom Persona Builder
+### Custom Personas
 
 ```bash
-# Interactive questionnaire (8 core traits)
-npx cbrowser persona-questionnaire start
+# Interactive questionnaire (13 core traits)
+cbrowser persona-questionnaire start
 
-# Comprehensive questionnaire (all 25 traits)
-npx cbrowser persona-questionnaire start --comprehensive --name "my-tester"
+# Comprehensive questionnaire (all 30 traits)
+cbrowser persona-questionnaire start --comprehensive --name "my-tester"
 
-# From a description — AI generates trait values
-npx cbrowser persona-create "A 68-year-old retiree who just got their first smartphone"
+# From a description
+cbrowser persona create "A 68-year-old retiree who just got their first smartphone" --name retiree
 ```
 
 ### Abandonment Detection
 
-The simulation stops when a realistic user would give up:
+A cognitive journey stops when a realistic user would give up:
 
-```bash
-⚠️ ABANDONED after 8 steps
+```text
+ABANDONED after 8 steps
 Reason: Patience depleted (0.08) - "This is taking too long..."
 Friction points:
   1. Password requirements unclear (step 4)
   2. Form validation error not visible (step 6)
 ```
 
+### Replay a Journey for Free
+
+Record a journey once with `--trace`, then replay it with no model calls. Replay is deterministic; a page that has changed since the recording stops the replay at the step that diverged.
+
+```bash
+cbrowser cognitive-journey --persona first-timer --start "https://your-site.com" --goal "sign up" --trace
+cbrowser cognitive-journey --persona first-timer --start "https://your-site.com" --goal "sign up" \
+  --replay-trace ~/.cbrowser/traces/<journeyId>.jsonl
+```
+
 ---
 
 ## Accessibility Empathy Testing
 
-Simulate how users with disabilities experience your site:
+Simulate how people with disabilities experience your site. Without `--goal` it checks the page for barriers and needs no key; with `--goal` it also runs a short journey per persona.
 
 ```bash
-npx cbrowser empathy-audit "https://your-site.com" \
-  --goal "complete signup" \
-  --disabilities "motor-impairment-tremor" \
+cbrowser empathy-audit "https://your-site.com" \
+  --personas motor-impairment-tremor,dyslexic-user \
+  --device iphone-16 \
   --html
 ```
 
+Persona names accept the short aliases too (`adhd`, `autism`, `tremor`, `dyslexia`, ...). `--disabilities` is the older spelling of `--personas`.
+
 ---
 
 ## Constitutional AI Safety
 
-AI agents need boundaries. CBrowser classifies every action by risk level:
+AI agents need boundaries. CBrowser classifies every action by risk:
 
 | Zone | Examples | Behavior |
 |------|----------|----------|
-| 🟢 **Green** | Navigate, read, screenshot | Auto-execute |
-| 🟡 **Yellow** | Click buttons, fill forms | Log and proceed |
-| 🔴 **Red** | Submit, delete, purchase | Requires verification |
-| ⬛ **Black** | Bypass auth, inject scripts | Never executes |
+| 🟢 **Green** | Navigate, read, screenshot | Runs |
+| 🟡 **Yellow** | Click buttons, fill forms | Logged and runs |
+| 🔴 **Red** | Submit, delete, purchase | Refused unless forced |
+| ⬛ **Black** | Bypass auth, inject scripts | Never runs |
+
+**How red is decided:** by the element an action would activate as well as by the selector text. The selector string is classified first, so naming a dangerous action in a selector is refused even when the element reads harmless. The element is then classified however it is addressed: the same "Delete account" button is refused by its label, its id, its test id or a structural path, and an element with only a delegated click listener is judged by its own text. A click on a container is judged at the point it lands (beneath fixed bars and overlays, into open shadow roots, slots and same-origin frames, up to the control, pointer-cursor item or text-bearing item there) and the click is sent to that judged point. A refused click is never retried with a "healed" selector.
 
-**How red is decided:** by the element an action would activate as well as by the selector text. The selector string is classified first, as before, so naming a dangerous action in a selector is refused even when the element reads harmless. The element is then classified however it is addressed: the same "Delete account" button is refused by its label, its id, its test id or a structural path, and an element with only a delegated click listener is judged by its own text. A click on a container is judged at the point it lands - beneath fixed bars and overlays, into open shadow roots, slots and same-origin frames, up to the control, pointer-cursor item or text-bearing item there - and the click is sent to that judged point. A refused click is never retried with a "healed" selector. Gated: `click`, `smart_click`, hover-click and the daemon's click; `press_key`, `type_text` and `keyboard` (Enter, including a typed or pressed newline, or Space on the focused control - followed into shadow roots and same-origin frames - and Enter in a field whose form submits one, judged together with a combobox's `aria-activedescendant` item); a `drag` whose press and release share a control; and the NL test runner's `click at X, Y`. Caller-supplied page script is red by itself, whatever it does: `evaluate_script` over MCP, `evaluate`/`eval` on the CLI and the daemon's evaluate are refused without force and run nothing, because a script can activate any control without passing the element gate (`document.querySelector('#delete-account').click()` is the click `click` refuses, in one line). Names below a control (icon labels, image alts, slotted and open-shadow text, `<label for>`) and the path a form submits to (a submitter's `formaction`, otherwise the form's `action`) feed the same keyword classification. `force: true` on the MCP `click`, `press_key`, `type_text`, `drag` and `evaluate_script` tools (`smart_click` has none: use `click`), or `--force` on the CLI, proceeds and reports the zone.
+**Gated:** `click`, `smart_click`, hover-click and the daemon's click; `press_key`, `type_text` and `keyboard` (Enter, including a typed or pressed newline, or Space on the focused control, followed into shadow roots and same-origin frames, and Enter in a field whose form submits one, judged together with a combobox's `aria-activedescendant` item); a `drag` whose press and release share a control; and the natural-language runner's `click at X, Y`. Caller-supplied page script is red by itself: `evaluate_script` over MCP, `evaluate`/`eval` on the CLI and the daemon's evaluate are refused without force and run nothing, because a script can activate any control without passing the element gate. Names below a control (icon labels, image alts, slotted and open-shadow text, `<label for>`) and the path a form submits to feed the same keyword classification. `force: true` on the MCP `click`, `smart_click`, `press_key`, `type_text`, `drag` and `evaluate_script` tools, or `--force` on the CLI, proceeds and reports the zone.
 
-Not judged, by design: clicking or typing into a text field or a contenteditable editor, and choosing a value with a native checkbox, radio or select or a `role=checkbox`, `radio` or `switch` - the submit that commits them is judged, but a switch or checkbox that saves on its own as it changes commits immediately and is not refused; and a link's address (a link to `/buy` is navigation). Not gated: cognitive journeys and other engines that choose their own clicks. Known limits: labels inside closed shadow roots or drawn by CSS, controls inside cross-origin frames, icon-only controls with no accessible name, a confirmation button with neutral text ("Confirm", "OK") inside a dialog, a form with no `action` or an empty `formaction` (it submits to the page's own address, which is not judged), an `aria-activedescendant` id that only resolves outside the field's own document or shadow root (not followed; ARIA references do not cross roots), a Tab inside typed text that moves focus before a later Enter, one round trip between the check and the click in which a page could swap the element, and keyword false positives: a "Purchase history" tab, a "Buyer protection" link, a `role=option` value picker such as "Pay with PayPal", a harmless "Apply" in a form posting to `/checkout/discount`, and an inert heading such as "Checkout" clicked by id (refusing it costs one forced retry; missing a listener-only "Delete account" cannot be undone). Classification uses English keyword patterns; when the element cannot be read, the selector text alone decides.
+**Not judged, by design:** clicking or typing into a text field or a contenteditable editor; choosing a value with a native checkbox, radio or select, or a `role=checkbox`, `radio` or `switch` (the submit that commits them is judged, but a switch that saves as it changes commits immediately); and a link's address (a link to `/buy` is navigation). **Not gated:** cognitive journeys and other engines that choose their own clicks.
+
+**Known limits:** labels inside closed shadow roots or drawn by CSS, controls inside cross-origin frames, icon-only controls with no accessible name, a confirmation button with neutral text ("Confirm", "OK") inside a dialog, a form with no `action` or an empty `formaction`, an `aria-activedescendant` id that only resolves outside the field's own root, a Tab inside typed text that moves focus before a later Enter, one round trip between the check and the click in which a page could swap the element, and keyword false positives such as a "Purchase history" tab, a "Buyer protection" link, a `role=option` picker such as "Pay with PayPal", or a harmless "Apply" in a form posting to `/checkout/discount`. Classification uses English keyword patterns; when the element cannot be read, the selector text alone decides.
 
 ---
 
-## Site Knowledge System
+## Browser Control from the CLI
+
+Each `cbrowser` command is its own process. The persistent profile keeps cookies and the last URL between commands, but not the page itself, so the next command reloads it. To keep one live page across commands (scroll position, open menus, form state), start the daemon:
+
+```bash
+cbrowser daemon start --device iphone-16 --timeout 45m
+cbrowser navigate "https://your-site.com"
+cbrowser scroll down 600
+cbrowser click "Pricing"
+cbrowser screenshot pricing.png
+cbrowser network list --filter api --json
+cbrowser console --filter error
+cbrowser daemon stop
+```
 
-CBrowser learns your site over time:
+While the daemon runs, `navigate`, `click`, `fill`, `hover`, `type`, `press`, `keyboard`, `screenshot`, `extract`, `run`, `scroll`, `evaluate`, `capture`, `har`, `network` and `console` use its browser. Commands that would open a second browser on its profile refuse until you stop it.
 
-- **Page Understanding** — DOM analysis, page type classification, available affordances
-- **Site Model Learning** — Persistent knowledge graph across sessions
-- **Cross-Session Profiles** — AES-256-GCM encrypted cookie/state profiles
-- **Goal Decomposition** — Sub-goal trees with fallback strategies
+Global browser flags work on every command that opens a browser: `--device <name>` (200+ presets; `cbrowser device list`), `--viewport 390x844` or a size name, `--browser chromium|firefox|webkit`, `--geo <preset or lat,lon>`, `--locale`, `--timezone`, `--no-headless` and `--no-persistent`. Flags also take `--flag=value`.
 
 ---
 
 ## Natural Language Testing
 
 ```txt
-# Test: Checkout Flow
-go to https://your-site.com/products
-click "Add to Cart" button
-verify page contains "1 item in cart"
-click checkout
-fill email with "test@example.com"
-click "Place Order"
-verify url contains "/confirmation"
+# Test: Search
+go to https://your-site.com
+type "lamps" in the search box
+click the search button
+verify page contains "results"
+verify url contains "/search"
 ```
 
 ```bash
-npx cbrowser test-suite checkout-test.txt --html
+cbrowser test-suite search-test.txt --html --output report.json
 
 # Auto-repair broken tests
-npx cbrowser repair-tests broken-test.txt --auto-apply --verify
+cbrowser repair-tests broken-test.txt --auto-apply --verify
 ```
 
+Steps that would submit, delete or buy (for example `click "Place Order"`) are red zone, and the test runner never forces them. Keep those out of unattended suites or test them against a sandbox account.
+
 ---
 
 ## Visual Testing
 
 ```bash
 # AI visual regression
-npx cbrowser ai-visual capture "https://your-site.com" --name homepage
-npx cbrowser ai-visual test "https://staging.your-site.com" homepage --html
+cbrowser ai-visual capture "https://your-site.com" --name homepage
+cbrowser ai-visual test "https://staging.your-site.com" homepage --html
 
-# Cross-browser (Chrome, Firefox, Safari)
-npx cbrowser cross-browser "https://your-site.com" --html
+# Cross-browser (Chromium, Firefox, WebKit)
+cbrowser cross-browser "https://your-site.com" --html
 
-# Responsive (mobile, tablet, desktop)
-npx cbrowser responsive "https://your-site.com" --html
+# Responsive (mobile, tablet, desktop by default)
+cbrowser responsive "https://your-site.com" --html
 
-# A/B comparison (staging vs production)
-npx cbrowser ab "https://staging.your-site.com" "https://your-site.com" --html
+# A/B comparison (staging against production)
+cbrowser ab "https://staging.your-site.com" "https://your-site.com" --html
 ```
 
 ---
 
-## Competitive UX Benchmark
+## Screen Capture
 
-Run identical cognitive journeys across your site and competitors:
+Record a GIF, animated WebP or video of a page, plus a frame manifest an AI can read:
 
 ```bash
-npx cbrowser competitive-benchmark \
-  --sites "https://your-site.com,https://competitor-a.com,https://competitor-b.com" \
-  --goal "sign up for free trial" \
-  --persona first-timer \
-  --html
+cbrowser capture start "https://your-site.com" --duration 5s --format gif,webp
+cbrowser capture start "https://your-site.com" --until-element-gone ".spinner" --timeout 20s
 ```
 
----
-
-## Upgrading to 19.0.0
-
-One breaking change: the agent-ready **accessibility** axis is now
-**agentPerceivability**, and three deprecated aliases are removed.
+With the daemon running, `capture start` records the daemon's page while you drive it with other commands, and `capture stop` finishes it.
 
-| removed | use instead |
-|---|---|
-| `score.accessibility` | `score.agentPerceivability` |
-| `summary.dynamicContentCount` | `summary.deferredLoadingPatterns` |
-| `comparison.bestAccessibility` | `comparison.bestAgentPerceivability` |
-| issue `category: "accessibility"` | `category: "agentPerceivability"` |
+---
 
-All three replacements have carried identical values since 18.82.0, so if you
-upgraded within that range you can switch keys before upgrading and take this
-release with no further change.
+## Competitive UX Benchmark
 
-The rename is not cosmetic. That axis measures whether a *machine* can perceive
-your elements — ARIA present, elements carry text. It is **not** a WCAG
-conformance score, and it returned 100 on a page where `empathy_audit` found two
-WCAG violations. Both numbers were right; calling one of them "accessibility" was
-not. Run `empathy_audit` for conformance, and expect the two to differ.
+Run identical cognitive journeys on your site and your competitors' (needs an Anthropic API key; one journey per site):
 
-On input, `category: "accessibility"` is still accepted and scores identically, so
-tools that *send* issues need not change immediately. Tools that *read* scores do.
+```bash
+cbrowser competitive-benchmark \
+  --sites "https://your-site.com,https://competitor-a.com,https://competitor-b.com" \
+  --goal "sign up for free trial" \
+  --persona first-timer \
+  --html
+```
 
 ---
 
 ## Geo Proxy
 
-Test from 10 global regions via residential proxies:
+Test from 10 global regions via residential proxies. Today `agent-ready-audit` is the command that routes through them:
 
 ```bash
-npx cbrowser cognitive-journey \
-  --start "https://your-site.com" \
-  --persona first-timer \
-  --goal "sign up" \
-  --geo-region uk
+cbrowser agent-ready-audit "https://your-site.com" --geo-region uk
 ```
 
 **Regions:** `us-west`, `us-east`, `us-central`, `uk`, `germany`, `france`, `japan`, `australia`, `canada`, `brazil`
 
-Geo routing needs your own residential-proxy credentials — the package ships the
-mechanism and no keys. Set `IPROYAL_USERNAME` / `IPROYAL_PASSWORD`, or place a
-`geo-proxy.json` at `~/.cbrowser/`. Without either, `--geo-region` fails loudly
-rather than silently running from your own IP.
-
-A `geo-proxy.json` may define additional regions; they merge over the ten above.
+Geo routing needs your own residential-proxy credentials; the package ships the mechanism and no keys. Set `IPROYAL_USERNAME` / `IPROYAL_PASSWORD`, or place a `geo-proxy.json` in `~/.cbrowser/`. Without either, `--geo-region` exits 1 rather than silently running from your own IP. A `geo-proxy.json` may define additional regions; they merge over the ten above.
 
-> **Corrected 2026-08-11.** This section previously advertised 12 regions
-> including `india` and `singapore`. Those are not in the shipped defaults and
-> resolve to nothing on a clean install — they existed only in a local config
-> file on one machine. Advertising them was a claim the package could not keep.
+For location without a proxy (the browser's reported geolocation, not its IP), use `--geo new-york` or `--geo 40.71,-74.00` on any command.
 
 ---
 
 ## MCP Server Integration
 
-CBrowser runs as an MCP server for Claude.ai, Claude Desktop, and Claude Code.
+CBrowser runs as an MCP server for Claude.ai, Claude Desktop and Claude Code.
 
-### Claude.ai (Easiest — No Install)
+### Claude.ai (no install)
 
-Add the MCP connector and optionally install the Claude.ai Skill:
-
-1. Go to [Customize → Connectors](https://claude.ai/customize/connectors) → "Add custom connector"
+1. Go to [Customize → Connectors](https://claude.ai/customize/connectors) and choose "Add custom connector".
 2. Paste: `https://demo.cbrowser.ai/mcp`
-3. Download the [Claude.ai Skill (.zip)](https://cbrowser.ai/downloads/cbrowser-claudeai.skill) — gives Claude context about tools, pricing, and workflows
-4. Go to Customize → Skills → Upload the zip
-
-> **Note:** The Claude.ai Skill is a lightweight knowledge file for the web interface. It is NOT the same as the Claude Code Skill (see below).
+3. Optionally download the [Claude.ai Skill (.zip)](https://cbrowser.ai/downloads/cbrowser-claudeai.skill), which gives Claude context about the tools and workflows, and upload it under Customize → Skills.
 
-### Claude Code Skill (CLI — Not for Claude.ai)
+The Claude.ai Skill is a small knowledge file for the web interface. It is not the Claude Code Skill below.
 
-For [Claude Code](https://claude.ai/claude-code) terminal users:
+### Claude Code Skill
 
 ```bash
 curl -fsSL https://raw.githubusercontent.com/alexandriashai/cbrowser/main/scripts/install-skill.sh | bash
-npm install -g cbrowser && npx playwright install
+npm install -g cbrowser
 ```
 
-The Claude Code Skill is a full CLI integration with workflow routing, TypeScript tools, persistent memory, and constitutional safety. See [Skill Installation Guide](https://cbrowser.ai/docs/claude-skill-installation) for details.
+See the [Skill Installation Guide](https://cbrowser.ai/docs/claude-skill-installation).
 
 ### Claude Desktop
 
@@ -409,50 +449,57 @@ The Claude Code Skill is a full CLI integration with workflow routing, TypeScrip
 }
 ```
 
-### 122 MCP Tools
+The local stdio server registers the same 130 tools as the hosted one, plus 7 that keep their state in the local process (`empathy_audit_init`, `empathy_audit_record_barrier`, `empathy_audit_complete_persona`, `empathy_audit_summarize`, `get_emotional_state`, `list_emotional_personas`, `trigger_emotional_event`). `cbrowser mcp-remote` runs your own HTTP server for remote clients.
+
+### 130 MCP Tools
 
 | Category | Tools | Count |
 |----------|-------|-------|
-| **Navigation** | `navigate`, `screenshot`, `scroll`, `extract`, `analyze_page`, `find_element_by_intent` | 6 |
-| **Interaction** | `click`, `smart_click`, `fill`, `dismiss_overlay` | 4 |
-| **Cognitive Core** | `cognitive_journey_init`, `cognitive_journey_update_state`, `cognitive_journey_autonomous`, `cognitive_effort`, `cognitive_load_estimate`, `site_cognitive_assessment` | 6 |
-| **Cognitive Transport** | `cognitive_distance`, `cognitive_interpolate`, `cognitive_coverage`, `transport_map` | 4 |
-| **Attention & Visual Story** | `attention_analysis`, `attention_compare`, `visual_cognitive_story` | 3 |
-| **Persona** | `list_cognitive_personas`, `compare_personas`, `persona_create_*`, `persona_questionnaire_*`, `persona_trait_lookup`, `persona_values_*` | 15 |
-| **Testing** | `nl_test_inline`, `nl_test_file`, `generate_tests`, `repair_test`, `detect_flaky_tests`, `coverage_map` | 6 |
-| **Visual Testing** | `visual_baseline`, `visual_regression`, `responsive_test`, `cross_browser_test`, `cross_browser_diff`, `ab_comparison` | 6 |
-| **Site Intelligence** | `page_understand`, `site_model_query`, `site_model_status`, `site_profile_*` | 6 |
-| **AI Friendliness** | `agent_ready_audit`, `ai_benchmark`, `webmcp_ready_audit`, `remediation_patches`, `llms_txt_generate`, `llms_txt_validate`, `llms_txt_diff`, `structured_data_suggest` | 8 |
-| **Analysis** | `hunt_bugs`, `chaos_test`, `competitive_benchmark`, `empathy_audit` | 4 |
-| **Performance** | `perf_baseline`, `perf_regression` | 2 |
-| **Marketing** *(Enterprise)* | `marketing_campaign_*`, `marketing_audience_discover`, `marketing_compete`, `marketing_funnel_analyze`, `marketing_influence_matrix`, `marketing_lever_analysis`, `marketing_personas_list`, `list_influence_patterns` | 11 |
-| **Security** | `security_audit`, `web_security_scan` | 2 |
-| **Stealth** *(Enterprise)* | `stealth_enable`, `stealth_disable`, `stealth_status`, `stealth_check`, `stealth_diagnose`, `cloudflare_detect`, `cloudflare_wait` | 7 |
-| **Session & Browser** | `save_session`, `load_session`, `delete_session`, `list_sessions`, `list_baselines`, `browser_health`, `browser_recover`, `reset_browser`, `heal_stats`, `status` | 10 |
-| **Utility** | `assert`, `ask_user`, `set_api_key`, `api_key_status`, `clear_api_key`, `get_api_key_prompt` | 6 |
-
-### Token Cost & Selective Loading
-
-122 MCP tools consume ~45,000 tokens in LLM context. For cost-sensitive use, load selectively:
+| **Navigation and extraction** | `navigate`, `list_devices`, `screenshot`, `extract` | 4 |
+| **Interaction** | `click`, `smart_click`, `fill`, `scroll`, `dismiss_overlay`, `hover`, `type_text`, `press_key`, `handle_dialog`, `upload_file`, `drag` | 11 |
+| **Browser state** | `evaluate_script`, `get_console_messages`, `get_network_requests`, `manage_cookies`, `manage_storage`, `manage_tabs` | 6 |
+| **Cognitive journeys** | `cognitive_journey_init`, `cognitive_journey_update_state`, `list_cognitive_personas` | 3 |
+| **Cognitive transport and comparison** | `cognitive_effort`, `cognitive_load_estimate`, `cognitive_distance`, `cognitive_coverage`, `cognitive_interpolate`, `compare_personas`, `compare_personas_init`, `compare_personas_complete` | 8 |
+| **Attention and visual testing** | `attention_analysis`, `attention_compare`, `transport_map`, `visual_baseline`, `visual_regression`, `cross_browser_test`, `cross_browser_diff`, `responsive_test`, `ab_comparison` | 9 |
+| **Audits** | `agent_ready_audit`, `empathy_audit`, `webmcp_ready_audit`, `competitive_benchmark`, `site_cognitive_assessment`, `visual_cognitive_story` | 6 |
+| **Analysis and bugs** | `analyze_page`, `generate_tests`, `find_element_by_intent`, `ai_benchmark`, `hunt_bugs`, `chaos_test` | 6 |
+| **Testing** | `nl_test_inline`, `nl_test_file`, `repair_test`, `detect_flaky_tests`, `coverage_map`, `assert` | 6 |
+| **Performance** | `perf_baseline`, `perf_regression`, `list_baselines` | 3 |
+| **Remediation and llms.txt** | `remediation_patches`, `llms_txt_generate`, `llms_txt_validate`, `llms_txt_diff`, `structured_data_suggest` | 5 |
+| **Site knowledge** | `page_understand`, `site_model_status`, `site_model_query`, `site_profile_list`, `site_profile_status`, `site_profile_delete`, `question_answer` | 7 |
+| **Personas and values** | `persona_create_*` (6), `persona_traits_list`, `persona_manager`, `persona_update`, `persona_delete`, `persona_lookup`, `persona_values_list`, `persona_values_lookup`, `persona_trait_lookup`, `persona_category_guidance`, `persona_questionnaire_get`, `persona_questionnaire_build`, `list_influence_patterns` | 18 |
+| **Capture and artifacts** | `capture_start`, `capture_stop`, `capture_status`, `journey_heatmap_gif`, `artifact_fetch` | 5 |
+| **Sessions and browser** | `save_session`, `load_session`, `list_sessions`, `delete_session`, `status`, `browser_health`, `browser_recover`, `reset_browser`, `heal_stats` | 9 |
+| **Security** | `security_audit` | 1 |
+| **Marketing** | `marketing_personas_list`, `marketing_campaign_create`, `marketing_campaign_run`, `marketing_campaign_report_result` | 4 |
+| **Utility** | `ask_user` | 1 |
+| **Enterprise** *(stubs in the public package)* | `set_api_key`, `clear_api_key`, `api_key_status`, `get_api_key_prompt`, `cognitive_journey_autonomous`, `marketing_influence_matrix`, `marketing_lever_analysis`, `marketing_funnel_analyze`, `marketing_compete`, `marketing_audience_discover`, `marketing_discover_status`, `stealth_*` (5), `cloudflare_detect`, `cloudflare_wait` | 18 |
+
+The 18 Enterprise tools are listed by the public package so clients can discover them; they run on [CBrowser Enterprise](#enterprise-edition) and answer with an upgrade notice elsewhere.
+
+### Token Cost and Selective Loading
+
+All 130 tool definitions take a large share of an LLM's context. To expose only what you need, register tool families yourself:
 
 ```typescript
 import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+import { CBrowser } from "cbrowser";
 import {
   registerNavigationTools,
   registerInteractionTools,
   registerExtractionTools,
 } from "cbrowser/mcp-tools";
 
+const browser = new CBrowser({ headless: true });
 const server = new McpServer({ name: "my-app", version: "1.0.0" });
-const context = { getBrowser: () => browser };
+const context = { getBrowser: async () => browser };
 
-// ~5,000 tokens instead of ~45,000
 registerNavigationTools(server, context);
 registerInteractionTools(server, context);
 registerExtractionTools(server, context);
 ```
 
-**Categories:** `navigation`, `interaction`, `extraction`, `assertion`, `analysis`, `session`, `healing`, `visualTesting`, `testing`, `bugAnalysis`, `personaComparison`, `cognitive`, `cognitiveTransport`, `attention`, `siteKnowledge`, `values`, `performance`, `audit`, `browserManagement`, `security`, `marketing`, `remediation`, `llmsTxt`.
+**Families:** `registerNavigationTools`, `registerInteractionTools`, `registerAdvancedInteractionTools`, `registerExtractionTools`, `registerBrowserStateTools`, `registerAssertionTools`, `registerAnalysisTools`, `registerAuditTools`, `registerBugAnalysisTools`, `registerCaptureTools`, `registerCognitiveTools`, `registerPersonaComparisonTools`, `registerVisualTestingTools`, `registerTestingTools`, `registerPerformanceTools`, `registerRemediationTools`, `registerLlmsTxtTools`, `registerSiteKnowledgeTools`, `registerValuesTools`, `registerPersonaLifecycleTools`, `registerPersonaCreationTools`, `registerSessionTools`, `registerHealingTools`, `registerBrowserManagementTools`, `registerSecurityTools`, `registerMarketingTools`, `registerGifTools`, `registerArtifactTools`, `registerAskUserTool`.
 
 ---
 
@@ -469,17 +516,20 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v4
-      - uses: alexandriashai/cbrowser@v18
+      - uses: alexandriashai/cbrowser@v20.0.0
         with:
-          test-file: tests/e2e/checkout.txt
-          sensitivity: strict
+          test-file: tests/e2e/search.txt
 ```
 
+The action installs cbrowser and the browsers, then runs one of: `test-file` (a natural-language suite, with a JSON and an HTML report), `url` (writes a page analysis), or `command` (any cbrowser command). Outputs land in `cbrowser-output/` and are uploaded as an artifact. Pin the action to a release tag.
+
 ### Docker
 
+There is no published image. Build one from the repository's `Dockerfile`:
+
 ```bash
-docker run --rm -v $(pwd)/tests:/work/tests ghcr.io/alexandriashai/cbrowser:latest \
-  test-suite tests/checkout.txt --html
+docker build -t cbrowser .
+docker run --rm -v "$(pwd)/tests:/work/tests" cbrowser test-suite tests/search.txt --html
 ```
 
 ---
@@ -487,16 +537,16 @@ docker run --rm -v $(pwd)/tests:/work/tests ghcr.io/alexandriashai/cbrowser:late
 ## API Reference
 
 ```typescript
-import { CBrowser } from 'cbrowser';
+import { CBrowser } from "cbrowser";
 
 const browser = new CBrowser({
   headless: true,
   persistent: true,
 });
 
-await browser.navigate('https://example.com');
+await browser.navigate("https://example.com");
 
-const result = await browser.smartClick('Sign In', {
+const result = await browser.smartClick("Sign In", {
   maxRetries: 3,
   minConfidence: 0.8,
 });
@@ -512,26 +562,38 @@ await browser.close();
 ### Modular Imports
 
 ```typescript
-import { runVisualRegression, runCrossBrowserTest } from 'cbrowser/visual';
-import { runNLTestSuite, detectFlakyTests, repairTest } from 'cbrowser/testing';
-import { huntBugs, runChaosTest, findElementByIntent } from 'cbrowser/analysis';
-import { capturePerformanceBaseline, detectPerformanceRegression } from 'cbrowser/performance';
+import { runVisualRegression, runCrossBrowserTest } from "cbrowser/visual";
+import { runNLTestSuite, detectFlakyTests, repairTest } from "cbrowser/testing";
+import { huntBugs, runChaosTest, findElementByIntent } from "cbrowser/analysis";
+import { capturePerformanceBaseline, detectPerformanceRegression } from "cbrowser/performance";
 ```
 
+Also exported: `cbrowser/mcp-tools`, `cbrowser/mcp-remote`, `cbrowser/values` and `cbrowser/stealth`.
+
 ---
 
 ## Configuration
 
+Environment variables override the config file, and command-line flags override both.
+
 | Variable | Default | Description |
 |----------|---------|-------------|
 | `CBROWSER_DATA_DIR` | `~/.cbrowser` | Data storage directory |
-| `CBROWSER_HEADLESS` | `true` | Run headless |
-| `CBROWSER_BROWSER` | `chromium` | Browser engine |
+| `CBROWSER_HEADLESS` | `true` | Set `false` to show the browser |
+| `CBROWSER_BROWSER` | `chromium` | `chromium`, `firefox` or `webkit` |
+| `CBROWSER_VIEWPORT_WIDTH` / `CBROWSER_VIEWPORT_HEIGHT` | `1920` / `1080` | Default page size |
+| `CBROWSER_DEVICE` | none | Default device preset |
+| `CBROWSER_LOCALE` / `CBROWSER_TIMEZONE` | the host's | Browser locale and timezone |
 | `CBROWSER_TIMEOUT` | `30000` | Default timeout (ms) |
-| `CBROWSER_PROXY` | — | Proxy URL |
+| `CBROWSER_PROXY` | none | Proxy URL (or `CBROWSER_PROXY_SERVER`, `_USERNAME`, `_PASSWORD`) |
+| `CBROWSER_RECORD_VIDEO` | `false` | Record video of browser sessions |
+| `NO_COLOR` | unset | Disable colour output |
+
+**Config file:** the first of `./.cbrowserrc.json`, `./.cbrowserrc`, `./cbrowser.config.json`, `<data dir>/config.json` or `~/.cbrowserrc.json`. Keys include `browser`, `headless`, `viewport` (`{ "width", "height" }`), `device`, `locale`, `timezone`, `timeout` and `recordVideo`.
 
 ```bash
-npx cbrowser config set-api-key  # Required for cognitive journeys
+cbrowser config set-api-key <key>   # for cognitive journeys
+cbrowser doctor                     # check Node, Playwright, the browsers and the key
 ```
 
 ---
@@ -542,38 +604,36 @@ npx cbrowser config set-api-key  # Required for cognitive journeys
 
 | Feature | Description |
 |---------|-------------|
-| **Marketing Suite** | Test which design/copy/UX patterns influence which buyer segments |
-| **8 Marketing Personas** | B2B (enterprise-buyer, startup-founder, procurement-manager, technical-evaluator) + Consumer (impulse-shopper, price-researcher, loyal-customer, skeptical-first-timer) |
-| **Influence Matrix** | Conversion effectiveness for variant × persona combinations |
-| **Lever Analysis** | Which psychological persuasion patterns work for each persona |
-| **Constitutional Stealth** | Full stealth measures for authorized penetration testing |
+| **Marketing Suite** | Test which design, copy and UX patterns influence which buyer segments |
+| **8 Marketing Personas** | B2B (enterprise-buyer, startup-founder, procurement-manager, technical-evaluator) and consumer (impulse-shopper, price-researcher, loyal-customer, skeptical-first-timer) |
+| **Influence Matrix** | Conversion effectiveness for each variant and persona combination |
+| **Lever Analysis** | Which persuasion patterns work for each persona |
+| **Constitutional Stealth** | Stealth measures for authorized penetration testing |
 | **Web Security Scan** | OWASP-based security scanning |
+| **Your own API key** | Per-account Anthropic keys for autonomous journeys |
 
-All 125 MCP tools included with no rate limits.
+Unlimited tool use, with no credits.
 
 ---
 
 ## License
 
-**MIT License** — Free and open source. See [LICENSE](LICENSE).
+**MIT License.** Free and open source. See [LICENSE](LICENSE).
 
 ---
 
 © 2026 Alexandria Eden · [cbrowser.ai](https://cbrowser.ai) · [alexandria.shai.eden@gmail.com](mailto:alexandria.shai.eden@gmail.com)
 
----
-
 ## Links
 
-- **[Documentation](https://cbrowser.ai/docs)** — Full docs, guides, API reference
+- **[Documentation](https://cbrowser.ai/docs)**: guides and API reference
 - [NPM Package](https://www.npmjs.com/package/cbrowser)
 - [GitHub](https://github.com/alexandriashai/cbrowser)
 - [Issue Tracker](https://github.com/alexandriashai/cbrowser/issues)
-- [Roadmap](https://cbrowser.ai/docs/Roadmap)
 
 ### Research
 
-- [Research Methodology](https://cbrowser.ai/docs/Research-Methodology) — How the 25 traits were selected and validated
-- [Trait Index](https://cbrowser.ai/docs/Trait-Index) — All cognitive traits with citations
-- [Bibliography](https://cbrowser.ai/docs/Bibliography) — Academic references
-- [Persona Index](https://cbrowser.ai/docs/Persona-Index) — All 17 cognitive + 8 marketing personas
+- [Research Methodology](https://cbrowser.ai/docs/Research-Methodology): how the traits were selected and validated
+- [Trait Index](https://cbrowser.ai/docs/Trait-Index): every cognitive trait with citations
+- [Bibliography](https://cbrowser.ai/docs/Bibliography): academic references
+- [Persona Index](https://cbrowser.ai/docs/Persona-Index): every cognitive and marketing persona