filter-rental-search▌
rent.591.com.tw/filter-rental-search-fq54br · updated May 21, 2026
MDX-style export adds YAML metadata + attribution linking explainx.ai and this canonical listing URL.
Apply user-supplied filters (region, district, price, room count, property type, area, amenities, sort) on rent.591.com.tw and return matching rental listings with title, price, layout, area, floor, address, MRT distance, and detail URL.
| name | filter-rental-search |
| title | 591租屋網 Filter Rental Search |
| description | >- Apply user-supplied filters (region, district, price, room count, property type, area, amenities, sort) on rent.591.com.tw and return matching rental listings with title, price, layout, area, floor, address, MRT distance, and detail URL. |
| website | rent.591.com.tw |
| category | real-estate |
| tags | - real-estate - rental - taiwan - '591' - search - filters - read-only |
| source | 'browserbase: agent-runtime 2026-05-20' |
| updated | '2026-05-20' |
| recommended_method | url-param |
| alternative_methods | - method: browser rationale: >- The filter UI is fully clickable via .item-info card extraction + label-text matching, but every UI click maps 1:1 to a URL query-string param — constructing the URL directly skips 5-10 click-then-snapshot turns. Use the click path only to discover an unknown filter slug, then prefer URL params for production. - method: api rationale: >- Direct API path /v3/web/rent/list observed in the SSR X-Server-Monitor header is 404 to external clients — only callable server-side from the Nuxt SSR layer. No public JSON endpoint exists for filtered listings; the SSR'd HTML is the canonical data surface. |
| verified | false |
| proxies | false |
591租屋網 — Filter Rental Search
Purpose
Apply user-supplied rental filters (city, district, price range, room count, property type, area, amenities, sort order) on rent.591.com.tw (Taiwan's largest rental listings site) and return the matching listings — title, price, layout, area (坪 / ping), floor, address, district, distance to nearest metro station, agent/owner, last-updated time, and the canonical detail-page URL. Read-only — never contacts landlords, never logs in.
When to Use
- "Find me 2-bedroom apartments in Taipei 大安區 / 信義區 between NT$15,000 and NT$25,000."
- "Cheapest studios (套房) near a metro station in 新北市 板橋區."
- Rental-market monitoring: hourly/daily polling for new listings (
sort=posttime_desc&other=newPost). - Cross-region rent comparison (region 1 = Taipei, 3 = New Taipei, 8 = Taichung, 17 = Kaohsiung).
- Anywhere you would otherwise scrape 591 with brittle CSS selectors — filter via URL params, then read
.item-infocards.
Workflow
The recommended path is URL-param filtering + DOM extraction — every filter the site UI exposes is also a query-string param, so a single browse open with the right URL replaces 5–10 click-then-snapshot turns. The page is server-rendered (Nuxt SSR) so .item-info cards are present in the initial DOM before any JS runs. Anti-bot is minimal — bare Browserbase sessions (no --proxies, no --verified) successfully load and render listings. Ship stealth/proxies only if you start seeing ratelimit pages.
1. Construct the filter URL
Base: https://rent.591.com.tw/list?{params}
| Param | Meaning | Syntax | Example |
|---|---|---|---|
region | City (required, single value) | integer | region=1 (台北市) |
section | District(s) within region | comma-list of integers | section=5,7 (大安+信義) |
kind | Rental type | comma-list of integers | kind=2,1 (studio + whole apt) |
price | Rent range NT$/month | MIN_MAX; comma-join multiple ranges | price=15000_25000 or price=0_5000,30000_40000 |
layout | Room count | comma-list of 1–4 | layout=2,3 (2房 or 3房) |
shape | Building shape | comma-list of integers | shape=1,3,4 (公寓+透天+別墅) |
acreage | Floor area (坪 / ping) | MIN_MAX | acreage=10_30 |
other | Amenity / status flags | comma-list of slug strings (see enum below) | other=near_subway,pet,lift |
sort | Result ordering | slug | sort=posttime_desc (newest first; default = "預設排序" = no sort param) |
region enum (Taiwan-wide; gaps 9, 16, 18, 20 are non-Taiwan codes that render an empty title):
| ID | Region | ID | Region | ID | Region |
|---|---|---|---|---|---|
| 1 | 台北市 (Taipei City) | 2 | 基隆市 (Keelung) | 3 | 新北市 (New Taipei) |
| 4 | 新竹市 (Hsinchu City) | 5 | 新竹縣 (Hsinchu County) | 6 | 桃園市 (Taoyuan) |
| 7 | 苗栗縣 (Miaoli) | 8 | 台中市 (Taichung) | 10 | 彰化縣 (Changhua) |
| 11 | 南投縣 (Nantou) | 12 | 嘉義市 (Chiayi City) | 13 | 嘉義縣 (Chiayi County) |
| 14 | 雲林縣 (Yunlin) | 15 | 台南市 (Tainan) | 17 | 高雄市 (Kaohsiung) |
| 19 | 屏東縣 (Pingtung) | 21 | 宜蘭縣 (Yilan) | 22 | 台東縣 (Taitung) |
| 23 | 花蓮縣 (Hualien) | 24 | 澎湖縣 (Penghu) | 25 | 金門縣 (Kinmen) |
kind enum (only 1–4, 8–10 are valid for rent.591.com.tw; 5/6/7/11 redirect to other 591 properties — store.591, office.591, factory listings, sale-only land — and lose the rent filter UI):
| ID | Type |
|---|---|
| 0 | 不限 — all rental types (default; equivalent to omitting kind) |
| 1 | 整層住家 — whole apartment |
| 2 | 獨立套房 — independent studio (own bath/kitchen) |
| 3 | 分租套房 — sublet studio (own bath, shared common area) |
| 4 | 雅房 — private bedroom with shared bath |
| 8 | 車位 — parking space |
| 9 | 住宅 — residential aggregate |
| 10 | 套房 — studio aggregate (covers 2 + 3) |
section enum (region-specific; always re-derive per region — IDs are not stable across regions and you cannot port a Taipei section=5 to a New Taipei query). For 台北市 (region 1), discovered:
| ID | District | ID | District | ID | District |
|---|---|---|---|---|---|
| 1 | 中正區 | 2 | 大同區 | 3 | 中山區 |
| 4 | 松山區 | 5 | 大安區 | 6 | 萬華區 |
| 7 | 信義區 | 8 | 士林區 | 9 | 北投區 |
| 10 | 內湖區 | 11 | 南港區 | 12 | 文山區 |
For other regions, derive the mapping by clicking each district checkbox once and reading the URL — see the click-and-read snippet in Browser fallback.
other enum (amenity / status flags, comma-joined):
| Slug | Label |
|---|---|
newPost | 新上架 (newly listed) |
near_subway | 近捷運 (near MRT/metro) |
pet | 可養寵物 (pet-friendly) |
cook | 可開伙 (cooking allowed) |
cartplace | 有車位 (parking included) |
lift | 有電梯 (elevator) |
balcony_1 | 有陽台 (balcony) |
lease | 可短期租賃 (short-term OK) |
social-housing | 社會住宅 (social housing) |
rental-subsidy | 租金補貼 (subsidy-eligible) |
elderly-friendly | 高齡友善 (elderly-friendly) |
tax-deductible | 可報稅 (tax-deductible) |
naturalization | 可入籍 (household-registration eligible) |
shape enum (partial — derived by multi-select reverse-mapping shape=1,3,4 → 公寓 + 透天厝 + 別墅):
| ID | Type |
|---|---|
| 1 | 公寓 (walk-up apartment) |
| 3 | 透天厝 (townhouse / single-family) |
| 4 | 別墅 (villa) |
(2 and 5+ exist but were not exhaustively probed; click-and-read the UI if you need them.)
2. Open the URL
SID=$(browse cloud sessions create --keep-alive | node -e "let s='';process.stdin.on('data',c=>s+=c).on('end',()=>process.stdout.write(JSON.parse(s).id))")
export BROWSE_SESSION="$SID"
URL="https://rent.591.com.tw/list?region=1&kind=2,1&price=15000_25000&layout=2&sort=posttime_desc"
browse open "$URL" --remote
browse wait load --remote
# Listings are SSR'd into the initial HTML; no extra wait usually needed.
# Add `browse wait timeout 2000` only if the popup tooltip is blocking your evals.
The site auto-rewrites legacy param names (multiPrice → price, multiRoom → layout, etc.) — don't rely on the auto-rewrite, just send the canonical names from the table above.
3. Extract the total count
browse eval --remote 'JSON.stringify({total: (document.body.innerText.match(/已為你找到\s*([0-9,]+)\s*間/) || [])[1]});'
# → { "total": "1,525" }
The total reflects the full filtered match across all pages (not just the 30 cards rendered on page 1).
4. Extract the listing cards
The first page of the SSR'd response renders 30 listing cards in .item-info. Each card has a child <a> whose href is the canonical detail URL.
browse eval --remote '
const total = (document.body.innerText.match(/已為你找到\s*([0-9,]+)\s*間/) || [])[1];
const cards = [...document.querySelectorAll(".item-info")];
const listings = cards.map(c => {
const lines = (c.innerText || "").split("\n").map(l=>l.trim()).filter(Boolean);
const url = c.href || c.querySelector("a")?.href;
const id = url?.match(/\/(\d{6,})$/)?.[1];
const priceMatch = c.innerText.match(/([0-9,]+)\s*元\/月/);
// Layout/area/floor: a single line like "獨立套房14.9坪6F/7F" or "整層住家2房1廳1衛20坪3F/5F"
const sizeLine = lines.find(l => /\d+(\.\d+)?坪/.test(l)) || "";
const ping = parseFloat((sizeLine.match(/(\d+(\.\d+)?)坪/) || [])[1]) || null;
const floor = (sizeLine.match(/([0-9頂樓加蓋]+F\/\d+F)/) || [])[1] || null;
const layoutStr = (sizeLine.match(/(\d+房\d*廳\d*衛|獨立套房|分租套房|雅房|整層住家|開放式|樓中樓)/) || [])[1] || null;
// Address: the line containing "區-" (district-street)
const address = lines.find(l => /區-/.test(l)) || null;
// Distance to MRT: line starting with "距"
const distance = lines.find(l => l.startsWith("距")) || null;
return { id, url, title: lines[0], layout: layoutStr, area_ping: ping, floor, address, distance, price_ntd: priceMatch ? +priceMatch[1].replace(/,/g, "") : null };
});
JSON.stringify({total, count: listings.length, listings});
'
The first line of c.innerText is the listing title; the last line containing the substring 元/月 is the price; one line near the bottom of the card has the agent/owner + last-updated stamp + "昨日 X 人瀏覽" view counter (skip if not needed).
5. Pagination (if needed)
To paginate beyond the first 30 cards, append &firstRow=N with N in steps of 30: firstRow=30, firstRow=60, etc. (Some 591 mirrors call this pg= or page=, but firstRow is the canonical for rent.591.) The total count from step 3 tells you how many pages to walk.
6. Release the session
browse cloud sessions update "$SID" --status REQUEST_RELEASE
Browser fallback (when you need a filter slug you don't have)
If the user asks for a filter you don't yet have a URL slug for (e.g., a section value in 台中市), open the bare filter page, click the checkbox, and read the URL:
browse open "https://rent.591.com.tw/list?region=8" --remote # Taichung
sleep 2
browse eval --remote '[...document.querySelectorAll("label")].find(l=>l.innerText?.trim()==="北屯區")?.click(); "ok";'
sleep 1
browse get url --remote # → ...§ion=NN
This same recipe works for any filter chip — kind, shape, other, etc. The Vue store mutation is what builds the URL, so the canonical slug always appears in window.location.search immediately after the click.
Site-Specific Gotchas
browse snapshotconsistently fails on this site with a Stagehand parser error (observed across 8 attempts in iter 1, 2 of testing). Usebrowse evalfor DOM extraction directly. The accessibility-tree path is not viable; the explicit DOM-query path is.- No serious anti-bot. Bare Browserbase sessions (no
--proxies, no--verified) successfully load filtered pages and render the full.item-infogrid. The metadata for this skill reflects that —verified: false, proxies: false. Add stealth only if you observe Cloudfront 403s during sustained polling. - A "RoboForm 密碼管理工具" tooltip popup overlays the right side of the viewport on first page load. It does NOT block DOM evaluation, but it WILL show up in screenshots. Dismiss programmatically before screenshotting:
[...document.querySelectorAll("*")].filter(e=>e.children.length===0 && e.innerText?.trim()==="我知道了,不再提示")[0]?.click(); - Param auto-rewrite. Sending
multiPrice=15000_25000&multiRoom=2&kind=2,1rewrites tokind=2,1&price=15000_25000&layout=2on the next render. Don't rely on legacy names — use the canonicalprice,layout,section,shape,acreage,other,sortlisted in the table. sectionIDs are region-scoped and NOT portable. A Taipeisection=5(大安區) is meaningless in another region. Always re-derivesectionIDs per region with the click-and-read fallback recipe.regioncodes 9, 16, 18, 20 render a generic "租屋 | 房屋出租" title with no city name — these slots are reserved (probably for outlying Taiwan municipalities not listed in the rental flow). Only use the regions in the enum table; treat anything else as undefined.kind5/6/7/11 redirect to sibling 591 properties (store.591for retail/storefront, separate office/factory/land sites) and lose the rental filter UI. The valid rental kinds are 0, 1, 2, 3, 4, 8, 9, 10.- First page returns 30 cards regardless of filter strictness (vs. typical 12–20 elsewhere). Page 2 starts at
firstRow=30. Don't assume Craigslist-style 360-card batches. - Bare
https://rent.591.com.tw/sets aurlJumpIpcookie and 301-redirects to/list?region=Nbased on the request IP (in our trace, region=1 from any non-Taiwan IP — Taipei is the global default). Passregion=explicitly to lock the city. The redirect chain is:/→/list→/list?region=1. - Sort param's "默認" (default) is the empty value — sending
sort=or omittingsortboth yield the site's blended-relevance ordering (which up-ranks "優選好屋" promoted listings). Usesort=posttime_descif you want true newest-first across the whole result set. - Listing IDs in
https://rent.591.com.tw/{id}are 7–8 digit integers, not slugs. The detail page itself lives atrent.591.com.tw/{id}with no slug, no path segments. - Total count is in localized text only. Use the regex
/已為你找到\s*([0-9,]+)\s*間/againstdocument.body.innerText— there is nodata-*attribute exposing the integer count cleanly. The count includes the comma thousand-separator and you must strip it before parsing. - Card innerText is multi-line and order-sensitive. Title is
lines[0]. Size/floor/type is the line matching/\d+坪/. Address is the line matching/區-/. Don't rely on line indices — match by content shape. - Direct API path tried and failed.
GET https://rent.591.com.tw/v3/web/rent/list?region=1(the path observed in the CloudfrontX-Server-Monitorheader) returns a 404 to external clients — it's only callable server-side from the SSR layer. Don't waste turns trying to find a JSON endpoint; the SSR'd HTML is the canonical data surface.
Expected Output
{
"success": true,
"filters_applied": {
"region": 1,
"region_name": "台北市",
"section": [5, 7],
"section_names": ["大安區", "信義區"],
"kind": [1, 2],
"kind_names": ["整層住家", "獨立套房"],
"price_min": 15000,
"price_max": 25000,
"layout": [2],
"other": ["near_subway", "lift"],
"sort": "posttime_desc"
},
"url": "https://rent.591.com.tw/list?region=1§ion=5,7&kind=2,1&price=15000_25000&layout=2&other=near_subway,lift&sort=posttime_desc",
"total_matches": 1525,
"page_size": 30,
"listings": [
{
"id": "21297268",
"url": "https://rent.591.com.tw/21297268",
"title": "中山站、電梯套房、台水電、獨洗、有陽台、代收垃圾",
"layout": "獨立套房",
"area_ping": 14.9,
"floor": "6F/7F",
"address": "中山區-林森北路259巷",
"distance": "距中山459公尺",
"price_ntd": 23500
}
],
"error_reasoning": null
}
Outcome shapes:
// Zero matches (filter combination too restrictive)
{ "success": true, "total_matches": 0, "listings": [], "url": "...", "filters_applied": {...} }
// kind redirected to a sibling property (e.g. user passed kind=5 / store)
{ "success": false, "reason": "kind_not_supported_by_rent_591", "redirected_to": "store.591.com.tw" }
// region code in the reserved-but-unused slots (9/16/18/20)
{ "success": false, "reason": "invalid_region", "region": 16 }
How to use filter-rental-search on Cursor
AI-first code editor with Composer
Prerequisites
Before installing skills in Cursor, ensure your development environment meets these requirements:
- ›Cursor installed and configured on your development machine
- ›Node.js version 16.0+ with npm package manager (verify with
node --version) - ›Active project directory or workspace where you want to add filter-rental-search
Execute installation command
Execute the skills CLI command in your project's root directory to begin installation:
The skills CLI fetches filter-rental-search from GitHub repository rent.591.com.tw/filter-rental-search-fq54br and configures it for Cursor.
Select Cursor when prompted
The CLI will show a list of available agents. Use arrow keys to navigate and space to select Cursor:
Verify installation
Confirm successful installation by checking the skill directory location:
Reload or restart Cursor to activate filter-rental-search. Access the skill through slash commands (e.g., /filter-rental-search) or your agent's skill management interface.
Security & Verification Notice
We perform automated surface-level scans (Gen AI Scanner, Socket, Snyk) during installation. These checks detect common vulnerabilities but do not guarantee complete security. Always review skill source code and verify the publisher's reputation before production use.
Skills execute code in your development environment. Always verify the publisher's identity, review recent commits, and test in isolated environments before production deployment.
List & Monetize Your Skill
Submit your Claude Code skill and start earning
Use Cases▌
Task Automation & Efficiency
Automate repetitive workflows and reduce manual effort
Example
Generate reports, summarize documents, draft communications
Save 3-5 hours per week on routine tasks
Knowledge Enhancement
Learn new skills, understand complex topics, get expert guidance
Example
Explain concepts, provide examples, suggest learning resources
Accelerate learning and skill development by 2x
Quality Improvement
Enhance output quality through reviews, suggestions, and refinements
Example
Review drafts, suggest improvements, catch errors
Improve work quality by 30-40% with less effort
Implementation Guide▌
Prerequisites
- ›Claude Desktop or compatible AI client with skill support
- ›Clear understanding of task or problem to solve
- ›Willingness to iterate and refine outputs
Time Estimate
15-45 minutes depending on use case complexity
Installation Steps
- 1.Install skill using provided installation command
- 2.Test with simple use case relevant to your work
- 3.Evaluate output quality and relevance
- 4.Iterate on prompts to improve results
- 5.Integrate into regular workflow if valuable
Common Pitfalls
- ⚠Expecting perfect results without iteration
- ⚠Not providing enough context in prompts
- ⚠Using skill for tasks outside its intended scope
- ⚠Accepting outputs without review and validation
Best Practices▌
✓ Do
- +Start with clear, specific prompts
- +Provide relevant context and constraints
- +Review and refine all outputs before using
- +Iterate to improve output quality
- +Document successful prompt patterns
✗ Don't
- −Don't use without understanding skill limitations
- −Don't skip validation of outputs
- −Don't share sensitive information in prompts
- −Don't expect skill to replace human judgment
💡 Pro Tips
- ★Be specific about desired format and style
- ★Ask for multiple options to choose from
- ★Request explanations to understand reasoning
- ★Combine AI efficiency with human expertise
When to Use This▌
✓ Use When
Use when skill capabilities match your task, clear ROI on time saved, and you can validate outputs. Best for repetitive tasks, learning, and quality improvement.
✗ Avoid When
Avoid when task requires deep expertise you can't validate, involves sensitive decisions, or when learning process is more valuable than speed of completion.
Learning Path▌
- 1Familiarize yourself with skill capabilities and limitations
- 2Start with low-risk, non-critical tasks
- 3Progress to more complex and valuable use cases
- 4Build expertise through regular use and experimentation
Discussion
Product Hunt–style comments (not star reviews)- No comments yet — start the thread.
Ratings
4.6★★★★★64 reviews- ★★★★★Ganesh Mohane· Dec 16, 2024
filter-rental-search has been reliable in day-to-day use. Documentation quality is above average for community skills.
- ★★★★★Meera Bansal· Dec 16, 2024
Solid pick for teams standardizing on skills: filter-rental-search is focused, and the summary matches what you get after install.
- ★★★★★Advait Lopez· Dec 16, 2024
filter-rental-search has been reliable in day-to-day use. Documentation quality is above average for community skills.
- ★★★★★Amina Malhotra· Dec 12, 2024
filter-rental-search reduced setup friction for our internal harness; good balance of opinion and flexibility.
- ★★★★★Yuki Abebe· Dec 12, 2024
I recommend filter-rental-search for anyone iterating fast on agent tooling; clear intent and a small, reviewable surface area.
- ★★★★★Advait Yang· Dec 12, 2024
Keeps context tight: filter-rental-search is the kind of skill you can hand to a new teammate without a long onboarding doc.
- ★★★★★Sakshi Patil· Nov 7, 2024
filter-rental-search reduced setup friction for our internal harness; good balance of opinion and flexibility.
- ★★★★★Aanya Smith· Nov 7, 2024
filter-rental-search reduced setup friction for our internal harness; good balance of opinion and flexibility.
- ★★★★★Kofi Dixit· Nov 3, 2024
filter-rental-search has been reliable in day-to-day use. Documentation quality is above average for community skills.
- ★★★★★Aisha Srinivasan· Nov 3, 2024
Useful defaults in filter-rental-search — fewer surprises than typical one-off scripts, and it plays nicely with `npx skills` flows.
showing 1-10 of 64