Government Session Agendas

Workflow

1. Start from agenda-keyword search.
2. Collect matching publications with dates and titles.
3. Open item pages for links to related materials.
4. Return timeline of agenda publications.

Human setup (when needed)

• If language/UI ambiguity blocks extraction, ask user to open the target publication in browser and share URL; then continue.

Quality checks

• Separate agenda announcements from post-meeting communications.
• Preserve publication timestamps exactly.

Access reality

• Public access type: UI page with direct downloadable files.
• Verification run: 2026-02-24.
• https://valitsus.ee/otsing?filters%5Btype%5D=Uudis&filters%5Bkeyword%5D=Istungi%20p%C3%A4evakord&sort=created&page=1 (HTTP 200, text/html;, file links detected: 1)
• https://valitsus.ee/en (HTTP 200, text/html;, file links detected: 1)

Request contract

• Use the listed primary endpoints as authoritative entry points.
• If API/query parameters are only visible in-browser, preserve exact request URL, params, and headers in output metadata.
• If endpoint is UI-only, document click path and extraction method used.

Output schema expectations

• Keep at least: source URL, retrieval timestamp, publication/update date (if available), title/record label, and extracted governance-relevant fields.
• Preserve original field names when present in downloadable/API output.

Limits and caveats

• Confirm whether data is open-download, UI-only, or authenticated before claiming full access.
• Separate narrative/guidance text from measurable records.

Verification hooks

• Validate endpoint reachability and content type before extraction.
• Validate that each extracted claim is linked to a concrete source URL.