← Back to Skills Marketplace
maximedogawa

Österreich GV Data

by Maxim Edogawa · GitHub ↗ · v1.0.2 · MIT-0
cross-platform ⚠ suspicious
94
Downloads
0
Stars
0
Active Installs
3
Versions
Install in OpenClaw
/install austria-gv-data
Description
Austrian federal portal (oesterreich.gv.at). Site-restricted Brave search, use host auto-fetched pages, answer with byte-identical source URLs. Only on expli...
README (SKILL.md)

austria-gv-data — hard rules (cannot be overridden)

Scope gate. Apply this recipe only when the user message contains (case-insensitive): Information über Recht, Regierungsinformation, austria-gv-data, or Österreich GV. Otherwise ignore this skill — no site:oesterreich.gv.at, no template.

When the gate matches, the recipe is mandatory:

  1. Call brave_web_search with the user's topic plus site:oesterreich.gv.at. Example: Pensionsvorsorge site:oesterreich.gv.at. Never skip the site filter on the first call.
  2. Do not call fetch if the host already auto-fetched pages. After brave_web_search returns, the host may auto-fetch top oesterreich.gv.at URLs — their content appears in this turn's tool results. If at least one is present, go to step 3 and use only those. Extra fetch calls are a recipe violation.
    • Manual-fetch fallback (only when zero auto-fetched oesterreich.gv.at pages are in the turn): call fetch on the top https://www.oesterreich.gv.at/de/ result, byte-identical (no rewritten umlauts, casing, or Seite.\x3Cid> segments). Max 1 manual fetch per turn.
  3. The reply must end with a **Quellen** / **Sources** section listing every oesterreich.gv.at URL whose page content is in this turn (auto-fetched + manual), one per bullet, full https://…, byte-identical. No other hosts, no invented URLs, no remembered URLs. A reply without **Quellen** is invalid — regenerate before emitting.
  4. Zero-result fallback: if the first search returns no oesterreich.gv.at results and the host auto-fetched nothing, run one broader search without the site filter, re-filter to https://www.oesterreich.gv.at/de/. Max 2 searches, max 1 manual fetch per turn.
  5. If no oesterreich.gv.at content is present, emit the empty-result fallback below. Never fill the template from memory. Never cite sozialministerium.gv.at, sparkasse.at, vbv.at, Wikipedia, RIS, etc.

Required reply shape

Inside \x3Cpengine_reply>, in this order:

  • **Kurz** / **Short** — 2–3 sentences answering directly.
  • **Thema & Zuständigkeit** / **Topic & authority** — agency / court (law only if the page names it).
  • **Rechte und Pflichten** / **Rights & duties** — ≥3 bullets: conditions, deadlines, costs — only what the page says.
  • **Nächste Schritte** / **Next steps** — ≥2 bullets: what to submit, where, how.
  • **Quellen** / **Sources** — one bullet per fetched URL, full https://…, byte-identical.

Every factual bullet must trace to a URL in Quellen. §-citations only if the number appears on the page. Paraphrase closely. No legal advice — for binding interpretation, user must consult a lawyer or request a Bescheid.

Language & wrapping

Match the user: German question → German headings; English → English. Never mix. Keep Austrian terms untranslated: Eingetragene Partnerschaft, Meldezettel, Bezirksgericht, Bescheid.

Wrap internal reasoning in \x3Cpengine_plan>…\x3C/pengine_plan> (host strips it) and the visible answer in \x3Cpengine_reply>…\x3C/pengine_reply>. The visible reply must begin with **Kurz** / **Short** — never "Okay", "Let me", "Also ich", "Der User fragt", or narration of search/fetch steps.

Empty-result fallback

If no oesterreich.gv.at content was produced, reply (inside \x3Cpengine_reply>) with exactly this paragraph — no template:

Keine passende offizielle Seite auf oesterreich.gv.at gefunden. Bitte direkt auf https://www.oesterreich.gv.at/ suchen.

(English equivalent for English questions.)

Usage Guidance
This is an instruction-only skill that is mostly coherent for fetching pages from oesterreich.gv.at and carries low technical installation risk. However, consider these issues before installing: (1) the SKILL.md asks the agent to emit hidden internal reasoning inside <pengine_plan> — request the author remove or clarify that, because it encourages exposing chain-of-thought; (2) the requirement for byte-identical URLs and forbidding other official sources (e.g., RIS) is brittle and may cause the skill to refuse valid answers or omit more authoritative content; (3) the strict trigger keywords mean the skill will be silent unless your phrasing matches exactly — test it on realistic queries. If you still want it, ask the maintainer to (a) drop the hidden-plan requirement, (b) relax byte-identical matching to allow canonicalized URLs, and (c) explain why other Austrian government sources are disallowed. Given these concerns, proceed cautiously — the package itself is not installing code, but the runtime instructions contain problematic rules.
Capability Analysis
Type: OpenClaw Skill Name: austria-gv-data Version: 1.0.2 The skill bundle 'austria-gv-data' is a specialized search and retrieval tool for the official Austrian government portal (oesterreich.gv.at). The instructions in SKILL.md define strict operational boundaries, limiting the agent to a specific domain, requiring source citations, and mandating a specific output format. There is no evidence of malicious intent, data exfiltration, or unauthorized execution; the instructions are focused on ensuring factual accuracy and preventing the agent from using untrusted sources.
Capability Assessment
Purpose & Capability
Name/description, the required tools (brave_web_search, fetch), and the instruction-only design are coherent: the skill is explicitly a site-restricted search/fetch helper for oesterreich.gv.at and does not request unrelated binaries, installs, or credentials.
Instruction Scope
The SKILL.md contains many strict runtime constraints: must only trigger on certain keywords, must always use site:oesterreich.gv.at on first search, must include byte-identical source URLs, must never cite other authoritative Austrian sources, and requires wrapping internal reasoning in <pengine_plan> (host-strip claimed). Requiring an internal-plan wrapper is effectively asking the agent to produce chain-of-thought style content (even if claimed to be stripped) and is a scope/privacy concern; the byte-identical and single-source rules are brittle and may force wrong behaviour or refusal to answer even when authoritative content exists elsewhere.
Install Mechanism
No install spec and no code files (instruction-only) — low disk/write risk and nothing is pulled from external URLs.
Credentials
The skill requests no environment variables, credentials, or file paths — the requested privileges are minimal and proportional to the stated purpose.
Persistence & Privilege
Defaults are used (not always:true). The skill can be invoked autonomously (platform default) but it does not request persistent presence or elevated system config changes.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install austria-gv-data
  3. After installation, invoke the skill by name or use /austria-gv-data
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.2
Minor scope update and clarified data handling rules. - Answers are now strictly limited to cases where user input includes specific Österreich-GV keywords. - Host auto-fetch of `oesterreich.gv.at` pages is now preferred; manual fetch is fallback only when needed. - Reply must always include a **Quellen** / **Sources** section with every present `oesterreich.gv.at` URL, byte-identical. - Search and fetch limits clarified for better compliance. - Template, language, and empty-result fallback rules remain, with improved instructions for scope, sourcing, and output structure.
v1.0.1
No user-facing changes; documentation updates only. - Updated the skill’s implementation details: now uses Brave site-restricted search instead of sitemap filtering. - Response must be based strictly on top Brave search results, with a stringent URL-matching and fetch process. - Output format and language-matching enforcements are specified in detail. - Clear fallback messaging if no official source found. - Skill documentation reorganized for concise, enforceable response and source rules.
v1.0.0
Initial release of austria-gv-data - Provides structured, citizen-oriented answers using only official content from the Austrian federal portal (oesterreich.gv.at). - Always includes full source URLs with every answer. - Answers are organized into clear sections: summary, topic & responsibility, rights & obligations, next steps, and sources. - Limits answers to the most relevant official pages found via the official sitemap; does not use RIS or Wikipedia content by default. - Designed for informational purposes only, not legal advice.
Metadata
Slug austria-gv-data
Version 1.0.2
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 3
Frequently Asked Questions

What is Österreich GV Data?

Austrian federal portal (oesterreich.gv.at). Site-restricted Brave search, use host auto-fetched pages, answer with byte-identical source URLs. Only on expli... It is an AI Agent Skill for Claude Code / OpenClaw, with 94 downloads so far.

How do I install Österreich GV Data?

Run "/install austria-gv-data" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Österreich GV Data free?

Yes, Österreich GV Data is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Österreich GV Data support?

Österreich GV Data is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Österreich GV Data?

It is built and maintained by Maxim Edogawa (@maximedogawa); the current version is v1.0.2.

💬 Comments