mcp: add detect_waf tool (28 -> 29); WAF-vendor identification + per-vendor strategy

Author: arham766

README.md

@@ -10,7 +10,7 @@ Stealth Chromium engine
 [![Chromium](https://img.shields.io/badge/chromium-151.0.7908.0-4285F4?logo=googlechrome&logoColor=white)](CHROMIUM_VERSION) [![pip](https://img.shields.io/badge/pip-3776AB?logo=pypi&logoColor=white)](https://pypi.org/project/tilion-fortress/) [![npm](https://img.shields.io/badge/npm-CB3837?logo=npm&logoColor=white)](https://www.npmjs.com/package/tilion-fortress) [![Docker pulls](https://img.shields.io/docker/pulls/tilion/fortress?logo=docker&logoColor=white&label=pulls)](https://hub.docker.com/r/tilion/fortress)<br/>
 [![CreepJS](https://img.shields.io/badge/CreepJS-0%25%20headless-2ea44f)](docs/GAUNTLET_RESULTS.md) [![Runtime.enable leak](https://img.shields.io/badge/Runtime.enable-no%20leak-2ea44f)](docs/GAUNTLET_RESULTS.md)<br/>
 [![Copy for agent](https://img.shields.io/badge/Copy%20for%20agent-24292f?logo=readme&logoColor=white)](https://raw.githubusercontent.com/tiliondev/fortress/main/AGENTS.md) [![llms.txt](https://img.shields.io/badge/llms.txt-24292f?logo=readme&logoColor=white)](https://raw.githubusercontent.com/tiliondev/fortress/main/llms.txt)<br/>
-[![MCP server](https://img.shields.io/badge/MCP-fortress%20·%2028%20tools-6E56CF?logo=modelcontextprotocol&logoColor=white)](mcp/) [![npm tilion-mcp](https://img.shields.io/npm/v/tilion-mcp?logo=npm&logoColor=white&label=npx%20tilion-mcp&color=CB3837)](https://www.npmjs.com/package/tilion-mcp)
+[![MCP server](https://img.shields.io/badge/MCP-fortress%20·%2029%20tools-6E56CF?logo=modelcontextprotocol&logoColor=white)](mcp/) [![npm tilion-mcp](https://img.shields.io/npm/v/tilion-mcp?logo=npm&logoColor=white&label=npx%20tilion-mcp&color=CB3837)](https://www.npmjs.com/package/tilion-mcp)
 
 **Fortress is a stealth Chromium engine that stops your scrapers and browser agents from getting blocked, with one line of code change.** Bot detectors flag automation by reading the browser fingerprint; Fortress corrects that fingerprint inside Chromium's C++, so the browser presents as an ordinary Chrome install. Scrapers finish their runs, agents reach the pages they were sent to, and CreepJS, Sannysoft, BrowserScan, and live Cloudflare Turnstile all read it as human. Point your existing Playwright or Puppeteer at Fortress over CDP, and nothing else in your code changes.
 
@@ -94,7 +94,7 @@ pip install -U tilion-fortress       # or:  docker run --rm -p 9222:9222 tilion/
 | | |
 |---|---|
 | **[What it is](#what-it-is)** · **[Quick start](#quick-start)** | what it is, install, first script, AI-agent setup |
-| **[The Fortress MCP](#the-fortress-mcp--stealth-browsing-as-agent-tools)** | 28 stealth-browser tools for AI agents (Beta) |
+| **[The Fortress MCP](#the-fortress-mcp--stealth-browsing-as-agent-tools)** | 29 stealth-browser tools for AI agents (Beta) |
 | **[Why patch the engine, not the page](#why-patch-the-engine-not-the-page)** | the self-revealing-JS thesis + the three detection layers |
 | **[How Fortress compares](#how-fortress-compares)** | vs puppeteer-stealth · Camoufox · CloakBrowser · closed vendors |
 | **[Proof: live-detector results](#proof-live-detector-results)** | CreepJS / Sannysoft / BrowserScan / Cloudflare, with screenshots |
@@ -273,7 +273,7 @@ Full guide: https://github.com/tiliondev/fortress/blob/main/AGENTS.md
 
 ## The Fortress MCP — stealth browsing as agent tools &nbsp;<sub>Beta</sub>
 
-Raw CDP is for code you write. The **Fortress MCP** is for agents that call **tools**: a [Model Context Protocol](https://modelcontextprotocol.io) server that hands Claude, Cursor, or any MCP client a stealth browser, so the moment a fetch is blocked it just calls a tool and gets the page. **28 tools, local and free** — `fetch_protected_page`, `extract_page`, `crawl_site`, `recon_site_apis`, `search_web`, `run_browser_task`, `save_profile`, `get_stealth_cdp_endpoint`, and more.
+Raw CDP is for code you write. The **Fortress MCP** is for agents that call **tools**: a [Model Context Protocol](https://modelcontextprotocol.io) server that hands Claude, Cursor, or any MCP client a stealth browser, so the moment a fetch is blocked it just calls a tool and gets the page. **29 tools, local and free** — `fetch_protected_page`, `extract_page`, `crawl_site`, `recon_site_apis`, `search_web`, `run_browser_task`, `save_profile`, `get_stealth_cdp_endpoint`, and more.
 
 <p align="center"><img src="mcp/demo.gif" alt="Same site, same prompt: a vanilla browser is blocked by PerimeterX while an agent with the Fortress MCP returns clean JSON" width="760"/></p>
 
@@ -324,7 +324,7 @@ Then just ask your agent — *“get the price off this StockX page”* — and
 
 Tools are annotated (reads auto-approve, writes gate), **pre-warmed** on startup (~100 ms first call), concurrency-safe, and timeout- and SSRF-guarded. A hosted endpoint with **residential egress is coming soon**.
 
-→ Full 28-tool table, benchmarks, and the agent skill: **[`mcp/`](mcp/README.md)**
+→ Full 29-tool table, benchmarks, and the agent skill: **[`mcp/`](mcp/README.md)**
 
 ---
 

mcp/README.md

@@ -1,6 +1,6 @@
 # Fortress MCP — a stealth browser for AI agents
 
-> **Beta** · 28 tools · runs **local & free** · **hosted cloud coming soon**
+> **Beta** · 29 tools · runs **local & free** · **hosted cloud coming soon**
 
 An [MCP](https://modelcontextprotocol.io) server that gives any AI agent the **Fortress
 stealth engine** the moment it gets blocked. When a fetch hits Cloudflare, DataDome,
@@ -85,7 +85,7 @@ server, then confirm with `get_egress_info`.
 
 If `tilion-mcp` isn't on PATH, use `"command": "python", "args": ["-m", "tilion.mcp"]`.
 
-## The 28 tools
+## The 29 tools
 
 | Tool | What the agent uses it for |
 |---|---|
@@ -98,6 +98,7 @@ If `tilion-mcp` isn't on PATH, use `"command": "python", "args": ["-m", "tilion.
 | `current_page` · `get_page_html` · `evaluate_js` · `wait_for` | inspect / script / wait on the working page |
 | `crawl_site` | crawl a whole site (auto-handles SPA/JS) → pages + sitemap |
 | `recon_site_apis` | reverse-engineer a site's private XHR/JSON API (secret-scrubbed) |
+| `detect_waf` | identify the anti-bot vendor (Cloudflare/DataDome/PerimeterX/Akamai/Kasada) + strategy |
 | `run_browser_task` · `list_browser_tasks` | 20 multi-step flows: login, paginate, infinite-scroll, checkout… |
 | `search_web` | web search through the stealth browser (no SERP API) |
 | `screenshot_page` · `save_page` · `download_file` | capture PNG / export pdf·html·text / download a file |

mcp/server.py

@@ -590,6 +590,24 @@ async def close_tab(index: int) -> dict:
     return await t.close_tab(int(index))
 
 
+@mcp.tool(annotations=_READ)
+@_safe
+async def detect_waf(url: str | None = None) -> dict:
+    """Identify the anti-bot / WAF vendor protecting a page (Cloudflare, DataDome,
+    PerimeterX, Akamai, Kasada, Incapsula) on the current page or a freshly-navigated `url`.
+
+    Returns {vendor, challenged, confidence, strategy, signals}. Use it to decide the
+    approach before hitting a hard target: `ip_and_wait`/Akamai and `behavior_*`/DataDome+
+    PerimeterX want residential egress + human interaction; `wait_challenge`/Cloudflare and
+    `pow_wait`/Kasada auto-clear or need time. `fetch_protected_page` already applies the
+    right strategy automatically and returns the same `waf` info.
+    """
+    if url:
+        await _check_url(url)
+    t = await _t()
+    return await t.detect_waf(url)
+
+
 @mcp.tool(annotations=_WRITE)
 @_safe
 async def solve_captcha(url: str | None = None) -> dict:
@@ -659,7 +677,7 @@ def main() -> None:
     # Banner to STDERR only — stdout is the MCP stdio transport and must stay clean.
     print(
         "Tilion Fortress Stealth-Browser MCP  [Beta]\n"
-        "  28 tools | local & free | hosted cloud (residential egress) coming soon\n"
+        "  29 tools | local & free | hosted cloud (residential egress) coming soon\n"
         "  docs: DOCUMENTATION.md - more coming; benchmarks in the README\n"
         "  listening on stdio...",
         file=sys.stderr, flush=True,