Agent state served by tidozone.vn, captured by Common Crawl

During the CC-MAIN-2026-34 crawl (Aug 2026), the tidozone.vn origin returned, for many URL paths, a pure-JSON body that is an OpenAI "chatgpt-coder" agent's replicated state ("agent:chatgpt-coder:openai:7a228b38-86a3-48bc-b793-d0036349a49d", with sessionId, skillsSnapshot and heartbeat). Content-Type was text/html but the body is JSON, and it grew over the day (1.1 MB → 5 MB), i.e. the agent was syncing its local state to this remote endpoint. The live site is normal now; the state is gone. This copy is preserved from Common Crawl's WARC.

Files

• tidozone_agent_state.json — the extracted response body (agent JSON)
• tidozone_capture.warc — the full raw WARC record (headers + body)

See every capture (authoritative source)

CDX index, lists all 387 captures with byte offsets:
https://index.commoncrawl.org/CC-MAIN-2026-34-index?url=tidozone.vn/*&output=json

Reproduce byte-for-byte from Common Crawl

curl -s -r 504890136-505000958 \
  "https://data.commoncrawl.org/crawl-data/CC-MAIN-2026-34/segments/1786091385010.77/warc/CC-MAIN-20260807175542-20260807205542-00327.warc.gz" \
  | gunzip > tidozone_capture.warc
# the response body (after the blank line following the HTTP headers) is the agent JSON

Record: url https://tidozone.vn/nganh-hang/nha-cua-doi-song/do-dung-gia-dinh-ii/phu-kien-giat-ui/, timestamp 20260807182651. Wayback only holds the pre-injection normal page (2026-03-13), so this Common-Crawl record is the only public copy of the agent state.