feat(adapter): add rescue flights adapter (El Al + Israir)

Adds @browserkit/adapter-rescue-flights — a public adapter that queries
real-time rescue flight availability from El Al and Israir.

El Al scraper:
- Scrapes the Angular CDK virtual-scroll seat availability page
- Collects flight data incrementally during scroll to handle DOM recycling
- Returns flight#, departure time, exact seat count for next 8 days

Israir scraper (Phase 1+2):
- Reads listing page DOM for all announced rescue flights
- Calls Israir's internal priceBar API per unique route for seat counts
  and prices — one API call per route instead of one browser nav per flight
- Intercepts FLIGHTS API responses via page.route() for exact flight#,
  departure time, and per-flight seat count on available dates
- All API calls via page.evaluate() (Imperva TLS-fingerprint bound)

Also adds:
- Makefile with agent-check target
- .claude/commands/agent-check.md — agent visual verification workflow
- Updated README, CLAUDE.md, AGENTS.md, progress.md, browserkit.config.js
- adapter-rescue-flights lives at jonzarecki/adapter-rescue-flights (not browserkit-dev org)

Made-with: Cursor

Author: jonzarecki

.claude/commands/agent-check.md

@@ -0,0 +1,70 @@
+# /agent-check
+
+Verify the rescue-flights adapter scrapers against the live sites.
+Use the cursor-ide-browser MCP tools directly for all browser interaction.
+Loop until all checks pass.
+
+## Tools to use
+
+- `browser_navigate` — load each live URL
+- `browser_take_screenshot` — capture the rendered page for visual comparison
+- `browser_snapshot` — get the ARIA tree for structural checks
+- `browser_scroll` — scroll the El Al list to expose virtual-scroll rows
+
+## Loop
+
+### Step 1 — Run scrapers
+
+```
+make agent-check
+```
+
+This builds the adapter, runs all four scraper calls, and writes
+`agent-check-results.json` to the repo root.
+
+Read the file. Note any entries where `availableSeats > 0` (El Al) and the
+full flight count (Israir).
+
+### Step 2 — El Al visual check (`to_israel`)
+
+1. `browser_navigate` → `https://www.elal.com/heb/seat-availability?d=1`
+2. Dismiss any cookie banner if present.
+3. Scroll down slowly (the list uses Angular CDK virtual scroll — rows only
+   render when in viewport). Take screenshots as you scroll.
+4. For each flight row visible in the screenshots, verify the scraper JSON
+   agrees on the seat count, especially any row showing a number > 0.
+
+**Failure condition:** a flight row visible on screen shows seats > 0 but the
+scraper JSON says 0 or is missing that entry entirely.
+
+### Step 3 — El Al visual check (`from_israel`)
+
+Same as Step 2 with `?d=0`.
+
+### Step 4 — Israir visual check (`to_tel_aviv`)
+
+1. `browser_navigate` → `https://www.israir.co.il/Flights/Rescue_Flights/To_Tel_Aviv`
+2. Take a screenshot. Count the visible flight cards.
+3. Use `browser_snapshot` to confirm button count.
+4. Verify scraper JSON count matches and all `origin` fields are IATA codes
+   (3 uppercase letters), not raw Hebrew text.
+
+### Step 5 — Israir visual check (`from_tel_aviv`)
+
+Same as Step 4 with the From_Tel_Aviv URL.  
+**Important:** this page loads flight cards asynchronously — wait for them to
+appear before counting. If the screenshot shows skeleton/placeholder UI, wait
+and retry the screenshot.
+
+### Step 6 — Evaluate
+
+- **All checks pass** → report PASS and stop.
+- **Any check fails** → identify the exact discrepancy, fix the relevant
+  source file in `packages/adapter-rescue-flights/src/`, then go back to Step 1.
+
+## Notes
+
+- Do not stop until the loop exits cleanly on a fresh run.
+- Transient page load failures are expected; just retry that step.
+- For El Al, the "show only flights with available seats" checkbox can be
+  ticked to reduce noise and make the screenshot easier to read.

.context/progress.md

@@ -1,5 +1,23 @@
 # Progress
 
+## Session 6 — Rescue Flights Adapter (2026-03-30)
+
+### `packages/adapter-rescue-flights` — new public adapter
+
+Built a complete Phase 1+2 rescue flights adapter covering El Al and Israir:
+
+- **El Al scraper** (`elal.ts`): Angular CDK virtual-scroll page. Scrolls in 300px steps and collects data at each position to handle DOM node recycling. Returns flight#, departure time, and exact seat count (0 = sold out) for all flights in the next 8 days.
+- **Israir scraper** (`israir.ts`): Two-phase approach.
+  - Phase 1: reads listing page DOM for all announced rescue flights (58+ cards, two directions).
+  - Phase 2: seeds a browser session, then calls Israir's internal `priceBar` API (`/api/results/priceBar`) per unique route for seat counts + prices across all dates. For available flights, intercepts the `FLIGHTS` API (`/api/search/FLIGHTS`) via `page.route()` to get exact flight#, departure time, and per-flight seat count.
+  - Total time: ~45s for from_tel_aviv (51 flights), ~15s for to_tel_aviv listing-only.
+  - All API calls via `page.evaluate()` — Imperva/TLS fingerprint bound, Node.js fetch rejected.
+- **`run-check.ts`** and **`Makefile`**: `make agent-check` calls adapter tools over MCP, dumps `agent-check-results.json`. Agent does visual verification using browser screenshots.
+- **14 unit tests** passing; integration tests updated for Phase 2 assertions.
+- **`browserkit.config.js`** updated: port 52746.
+- **`pnpm-workspace.yaml`** updated.
+- **README.md** updated with rescue flights entry in Available Adapters table.
+
 ## Session 0 — Project Setup (2026-03-22)
 
 - Synthesized SPEC.md from ChatGPT brainstorm conversation about authenticated MCP wrappers

.cursor/hooks/state/continual-learning-index.json

@@ -53,21 +53,25 @@
       "mtimeMs": 1774450242000,
       "lastProcessedAt": "2026-03-25T16:00:00.000Z"
     },
-    "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/672ffd14-9bc9-4c91-9eb7-9759b40e8893/672ffd14-9bc9-4c91-9eb7-9759b40e8893.jsonl": {
-      "mtimeMs": 1774522118000,
-      "lastProcessedAt": "2026-03-26T00:00:00.000Z"
-    },
     "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/d03e9a64-5187-4bee-88df-279763fa977e/d03e9a64-5187-4bee-88df-279763fa977e.jsonl": {
       "mtimeMs": 1774474691000,
       "lastProcessedAt": "2026-03-26T00:00:00.000Z"
     },
+    "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/672ffd14-9bc9-4c91-9eb7-9759b40e8893/672ffd14-9bc9-4c91-9eb7-9759b40e8893.jsonl": {
+      "mtimeMs": 1774790896000,
+      "lastProcessedAt": "2026-03-30T13:25:00.000Z"
+    },
     "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/8511b06d-07c8-42f2-8f30-5b6762c2fdc6/8511b06d-07c8-42f2-8f30-5b6762c2fdc6.jsonl": {
-      "mtimeMs": 1774523258000,
-      "lastProcessedAt": "2026-03-26T00:00:00.000Z"
+      "mtimeMs": 1774771011000,
+      "lastProcessedAt": "2026-03-30T13:25:00.000Z"
     },
     "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/9bec9e6f-a179-401c-90cc-6d64720fbfee/9bec9e6f-a179-401c-90cc-6d64720fbfee.jsonl": {
-      "mtimeMs": 1774534587000,
-      "lastProcessedAt": "2026-03-26T00:00:00.000Z"
+      "mtimeMs": 1774769808000,
+      "lastProcessedAt": "2026-03-30T13:25:00.000Z"
+    },
+    "/Users/jzarecki/.cursor/projects/Users-jzarecki-Projects-session-mcp/agent-transcripts/9e65043f-de65-4412-8de4-cb92e2d2d3be/9e65043f-de65-4412-8de4-cb92e2d2d3be.jsonl": {
+      "mtimeMs": 1774901830000,
+      "lastProcessedAt": "2026-03-30T15:00:00.000Z"
     }
   }
 }

.cursor/hooks/state/continual-learning.json

@@ -1,8 +1,8 @@
 {
   "version": 1,
-  "lastRunAtMs": 1774534549568,
-  "turnsSinceLastRun": 6,
-  "lastTranscriptMtimeMs": 1774534549508,
-  "lastProcessedGenerationId": "785251f5-f6dd-46fd-bd02-6b91246cd354",
+  "lastRunAtMs": 1774901809573,
+  "turnsSinceLastRun": 3,
+  "lastTranscriptMtimeMs": 1774901809436,
+  "lastProcessedGenerationId": "3000fdf6-5678-4931-8b00-e64206fb26f8",
   "trialStartedAtMs": null
 }

.gitignore

@@ -13,6 +13,9 @@ __pycache__/
 *.pyc
 .venv/
 
+# agent-check output — runtime, not for commits
+agent-check-results.json
+
 # OS
 .DS_Store
 

AGENTS.md

@@ -30,7 +30,7 @@ Durable facts and correction patterns for this workspace. Updated by continual-l
 - Browser mode switching (`headless` / `watch` / `paused`), screenshot, page state, and navigate are consolidated into a single `browser` MCP tool with an `action` parameter — user explicitly asked to reduce tool count ("too many tools"); do NOT revert to 5 separate management tools
 - Management tools bypass the LockManager; regular automation tools go through it
 - "Raw" Playwright access means exposing the CDP WebSocket URL (`wsEndpoint()`) of each adapter's browser — external agents (Claude Code, Cursor) attach to the already-authenticated session and write their own Playwright scripts via shell
-- The Playwright skill pattern: AI writes a script to `/tmp`, executes it via shell — primary approach for shell-capable clients (Cursor, Claude Code); opt-in `run_script` MCP tool planned for clients without shell access (Claude Desktop)
+- The Playwright skill pattern: AI writes a script to `/tmp`, executes it via shell — primary approach for shell-capable clients (Cursor, Claude Code); `run_script` MCP tool **implemented** using `quickjs-emscripten` (sandboxed QuickJS) for clients without shell access (Claude Desktop)
 - MCP resources use `page://${site}/snapshot` (site name dynamic) — user pushed back when the URI appeared to hardcode the adapter name
 - Testing utilities (`createTestAdapterServer`, `createTestMcpClient`) live at `@browserkit/core/testing` subpath — a separate harness package was explicitly rejected ("I don't think we need it, it should be in either adapter or in core")
 - Real Chrome (`channel: "chrome"`) is required for Google-based adapters — Playwright's bundled Chromium is blocked by Google's login with "This browser or app may not be secure". `isLoggedIn` must NOT navigate during login polling or it redirects the user away from the sign-in page.
@@ -41,7 +41,8 @@ Durable facts and correction patterns for this workspace. Updated by continual-l
 - Framework navigates to `adapter.loginUrl` before calling `isLoggedIn()` when browser is at `about:blank` — adapters do NOT need to handle this themselves
 - `warm_up_browser()` (visiting google/wiki/github before login) was evaluated from stickerdaniel's code — decided as "nice to have" for first-time login, not adopted yet
 - `browserkit login <site>` is blocked by the `CI=1` env var that Cursor sets — must run as `CI="" node packages/core/dist/cli.js login <site>` to open a headed browser from within Cursor terminal
-- `browser` tool `snapshot` action is planned — returns incremental aria-snapshot diff, more token-efficient than screenshots; `page-snapshot` MCP resource already exists, the action adds diff support. Inspired by `SawyerHood/dev-browser`.
+- `browser` tool `snapshot` action is **implemented** — returns incremental aria-snapshot diff (`track: boolean` returns only lines changed since last call via set-difference); `page-snapshot` MCP resource shares the same `captureSnapshot()` helper. Inspired by `SawyerHood/dev-browser`.
+- `browserkit login --connect` flag is **implemented** — attaches to an existing CDP endpoint via `autoDiscoverCdpEndpoint()` in `session-manager.ts`; useful when Chrome is already running.
 
 ## Design Process Preferences
 
@@ -61,6 +62,14 @@ Durable facts and correction patterns for this workspace. Updated by continual-l
 - Adapter roadmap priority: Reddit → Twitter/X (flagship, hardest bot detection) → Amazon (no consumer API at all) → Airbnb, Google Maps, Booking.com — all documented in the main README as planned adapters.
 - Reddit adapter is two-phase: Phase 1 = unauthenticated `old.reddit.com` (stable HTML class names, no login required); Phase 2 = authenticated (separate plan file). Target `old.reddit.com` exclusively — new Reddit is a React SPA with aggressive DOM churn.
 - Booking.com adapter was started (plan + Phase 1 architecture); user preference is to plan architecture first, then implement.
+- Booking.com adapter has **6 tools**: `get_upcoming_bookings`, `get_past_bookings`, `get_booking_details` (account tools) + `search_hotels`, `get_property`, `get_availability` (search/property tools). All 6 work in **headless mode** via CloakBrowser — no watch mode required.
+- **CloakBrowser** (`cloakbrowser` npm package) is in `@browserkit/core` `optionalDependencies` — lazily imported only when `antiDetection: { useCloakBrowser: true }` is set in adapter config. Provides 33 C++-level stealth patches that bypass DataDome. Do NOT use `channel: "chrome"` for Booking.com; use `useCloakBrowser: true` instead.
+- Global anti-detection patches applied to **all adapters** (unconditionally): `viewport: null` (avoids 1280×720 bot signal), `--blink-settings=primaryPointerType=4,...` (declares mouse at Blink level), `--window-size=1920,1080` (non-zero headless dimensions).
 - Phased adapter development pattern: Phase 1 is unauthenticated/mock, Phase 2 is authenticated/live. Add verification gates between phases.
 - Live scraping tests should run in GitHub CI via an external browser service (desired; specific service not yet chosen).
 - The main browserkit README doubles as the project's public-facing "blogpost" — user refers to it interchangeably; keep it polished and up-to-date with available + planned adapters
+- `adapter-rescue-flights` **has been implemented** (Phase 1 complete): scrapes El Al (`elal.com/heb/seat-availability?d=0`/`d=1`) via Angular DOM (`.seat-availability-list > .flight-group > .flight-item`) and Israir (`israir.co.il/Flights/Rescue_Flights/To_Tel_Aviv` / `/From_Tel_Aviv`) via internal APIs; both sites are public/unauthenticated.
+- Israir listing page shows flights as `button.flight-promotion.manual` elements — buttons have **no URLs and no data attributes**; clicking them opens a SPA booking widget, NOT a new page. Do NOT try to extract URLs from them.
+- Israir internal APIs discovered: `GET /api/results/priceBar?origin=TLV&destination=LCA&isOneWay=false&adults=1` (per-date seat counts + cheapest price) and `POST /api/search/FLIGHTS?origin=...&destination=...&startDate=...` with body `{"ignoredResults":0,"siteId":"isra2023"}` (flight number, departure/arrival times, exact seat count per segment).
+- Israir APIs require Imperva/Incapsula session cookies (`visid_incap_*`, `incap_ses_*`, `reese84`) bound to the browser's TLS fingerprint — Node.js `fetch` with copied cookies is rejected. Must call via `page.evaluate()` from within the browser context.
+- Israir scraping pattern (45s total vs 4 min naive): (1) read listing page DOM for all `button.flight-promotion.manual` cards, (2) seed-navigate to one search URL to establish Imperva session, (3) call `priceBar` API for all 6 destination cities in parallel via `page.evaluate()`, (4) intercept `POST /api/search/FLIGHTS` via route interception on 4 concurrent browser context pages for available dates only.

CLAUDE.md

@@ -35,6 +35,20 @@ TBD — stack not decided yet (TypeScript + Playwright or Python + Playwright).
 - Run `/plan` to decide what to work on next
 - Run `/review` before committing
 
+## Adapter verification (rescue-flights)
+
+After **any** change to `packages/adapter-rescue-flights/src/`, run the `/agent-check`
+command.  The agent will:
+
+1. Run `make agent-check` — builds the adapter and dumps scraper output to
+   `agent-check-results.json`.
+2. Open each live URL in the browser, take screenshots, and visually compare the
+   screenshots against the scraper JSON.
+3. Fix any discrepancy in the scraper source, then repeat from step 1.
+
+**Do not stop or commit until the visual comparison passes on all four checks.**
+The full workflow is documented in `.claude/commands/agent-check.md`.
+
 ## Forbidden patterns
 
 - No `any` types (TS) or untyped functions (Python)

Makefile

@@ -0,0 +1,20 @@
+PKG_DIR := $(dir $(abspath $(lastword $(MAKEFILE_LIST))))
+NODE    := node
+PNPM    := pnpm
+
+.PHONY: build test agent-check
+
+build:
+	$(PNPM) --filter @browserkit/adapter-rescue-flights build
+
+test: build
+	$(PNPM) --filter @browserkit/adapter-rescue-flights test
+
+## Call adapter tools over MCP HTTP (port 52746), dump results to agent-check-results.json.
+## Requires: browserkit start (adapter must already be running).
+agent-check: build
+	@echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
+	@echo "  Rescue Flights — agent-check"
+	@echo "  Adapter must be running: browserkit start"
+	@echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
+	$(NODE) $(PKG_DIR)packages/adapter-rescue-flights/dist/run-check.js

README.md

@@ -49,6 +49,7 @@ Configure your MCP client (Cursor, Claude Desktop, etc.):
 | [`@browserkit/adapter-hackernews`](https://github.com/browserkit-dev/adapter-hackernews) | Hacker News | none | `get_top`, `get_new`, `get_ask`, `get_show`, `get_comments` |
 | [`@browserkit/adapter-linkedin`](https://github.com/browserkit-dev/adapter-linkedin) | LinkedIn | required | `get_person_profile`, `get_company_profile`, `get_company_posts`, `search_people`, `search_jobs`, `get_job_details`, `get_feed` |
 | [`@browserkit/adapter-reddit`](https://github.com/browserkit-dev/adapter-reddit) | Reddit | none | `get_subreddit`, `get_thread`, `search`, `get_user` |
+| `@browserkit/adapter-rescue-flights` ([jonzarecki](https://github.com/jonzarecki/adapter-rescue-flights)) | El Al + Israir | none | `get_elal_flights`, `get_israir_flights` |
 
 ---
 

browserkit.config.js

@@ -11,6 +11,7 @@ export default {
       channel: "chrome",
     },
     "/Users/jzarecki/Projects/session-mcp/packages/adapter-linkedin/dist/index.js": { port: 52744, channel: "chrome" },
+    "@browserkit/adapter-rescue-flights": { port: 52746 },
     "/Users/jzarecki/Projects/browserkit-adapter-booking/dist/index.js": {
       port: 52745,
       // channel: "chrome" not used — CloakBrowser uses its own Chromium binary