Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Naver Real Estate Search
v0.6.3Search, compare, and monitor 대한민국 property listings from 네이버 부동산 with natural-language queries. Use when the user wants 강남 아파트 전세 시세 찾기, 특정 지역 매매/전세/월세 비교, 조...
⭐ 0· 131·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
The name/description (Naver property search/compare/watch) align with the included scripts and JSON data: search_real_estate.py, chat_real_estate.py, watch_real_estate.py, and candidate seed tooling. However, the SKILL metadata claims 'instruction-only' / no install spec while the package contains multiple runnable scripts and data files. The SKILL.md and several reference files also mention reusing logic from a local 'tmp/naverland-scrapper' upstream clone (classes such as src.core.parser.NaverURLParser), but the registry entry does not declare that dependency or any config path. This undeclared dependency and the mismatch between 'instruction-only' and the presence of code is an inconsistency worth noting.
Instruction Scope
The runtime instructions direct the agent/operator to run the included Python scripts (search/build/apply/chat/watch). Those scripts perform natural-language parsing, HTML-based candidate extraction (search.naver.com), direct new.land.naver.com complex detail fetches, and local file read/write (data/*.json, references/*.json). All of this is consistent with the stated purpose. The instructions do not ask for unrelated system files or credentials. They do, however, rely on network calls to Naver (search.naver.com and new.land.naver.com) and on local candidate-seed/candidate-cache files; the tool also documents handling 403/429 rate limits and storing blocked reasons. The agent will therefore perform outbound HTTP requests and update local JSON state — expected for a scraper but worth awareness.
Install Mechanism
There is no install spec (no packaged installation steps), which reduces installer risk. But the package does include multiple executable Python scripts and data files. Because files are bundled, invoking the skill will execute code shipped with it; this is not the same as an installer downloading external archives. The design mentions possible Playwright usage as a future improvement (in design.md), but there is no automatic download/install step in the current package. Still, the package references an external local upstream clone (tmp/naverland-scrapper) in documentation — if the runtime code expects that repo or other third-party components to be present, missing dependencies may cause the code to import or call out in unexpected ways when run.
Credentials
The skill declares no required environment variables, no credentials, and no special config paths. That is proportionate for a web-scraper that uses public web pages. The scripts read/write local JSON files inside the skill directory (data/, references/). There are no requests for unrelated tokens or secrets. One caution: references/candidate-seeds.generated.json contains an informational path pointing to a user's local workspace (C:\Users\...), but that is metadata from generation and not a declared required path; still confirm the code does not attempt to read arbitrary user-home files at runtime.
Persistence & Privilege
The skill does not request 'always: true' and does not declare actions to modify other skills or system-wide settings. It will persist state in its own data files (data/candidate-cache.json, data/watch-rules.json, references/candidate-seeds.json) inside the skill package's directory — this is expected for a watch/seed tool. Autonomous invocation (disable-model-invocation false) is normal and not a standalone red flag here.
What to consider before installing
What to check before you install or run this skill:
- Understand what it does: this package is a Naver (new.land.naver.com / search.naver.com) scraper and monitoring toolkit that makes outbound HTTP requests and stores state in local JSON files (data/ and references/). That behavior matches the description, but it will generate network traffic to Naver and write files to the skill folder.
- Review the code before running: although the SKILL.md presents it as instruction-only, the package contains runnable Python scripts. Inspect search_real_estate.py and any helper modules for network calls, third-party imports, or subprocess invocations you are not comfortable running. Confirm there are no unexpected remote endpoints (the visible endpoints are search.naver.com and new.land.naver.com, which are expected).
- Check dependencies: the docs reference a local upstream clone (tmp/naverland-scrapper) and reuse of parser/utility classes. Confirm whether the runtime imports actually require that external repo; if so, the skill may fail or attempt to import missing modules. If any external tooling (Playwright or headless browser) is added later, that increases install/runtime risk — run in an isolated environment.
- Rate limits and legality: the tool scrapes a third-party site and has logic for 403/429. Make sure you are comfortable with the site's terms of service and your organization’s policy about scraping before using it at scale.
- Run safely: run the scripts in a sandbox/VM or virtualenv first. Back up any existing local skill data you care about. If you only want the parser, use --parse-only/self-test modes rather than bulk scans. If you will use the watch functionality, be aware it writes to watch-rules.json and candidate-cache.json.
- If you plan to deploy it for automated use: add explicit monitoring, limit scraping frequency, and consider adding credentials/configuration for controlled notification channels (Telegram integration is mentioned but not built-in).Like a lobster shell, security has layers — review code before you run it.
latestvk97f6ac872xdex4ppvxedcmtx5839bqz
License
MIT-0
Free to use, modify, and redistribute. No attribution required.