=== Post-crawl corpus/documents verification (fresh live capture) === Captured: 2026-08-19, hours after the connector/pipeline test in Input 7 (the original test pipeline itself was deleted immediately after that test, per the "What worked" note — this capture re-confirms account-wide state independently, well after the fact). Endpoint: GET https://api.vectara.io/v2/corpora (account-wide corpora list) + GET https://api.vectara.io/v2/corpora/rag-eval-corpus-api/documents (documents in the scored corpus) --- Finding 1: no new corpus was created to receive crawled content --- The account contains exactly 5 corpora, all pre-existing and accounted for: - black-holes-sample-data (1 doc — unrelated Vectara sample data, present before this eval began) - semiconductor-sample-data (1 doc — unrelated Vectara sample data, present before this eval began) - rag-eval-corpus (11 docs — the earlier console-UI-driven ingestion pass) - rag-eval-corpus-api (11 docs, 10,072 parts — the scored corpus, API-only ingestion) - rag-eval-tenant2 (2 docs — the isolation-test corpus) No "web-crawl", "pipeline", or similarly-named corpus exists. If the connector run had written its crawled pages anywhere, a 6th corpus or a doc-count increase on one of the 5 above is the only place it could show up — neither happened. --- Finding 2: rag-eval-corpus-api still contains exactly the original 11 documents --- Full document ID list (paginated, 2 pages): 1. 01_tables_sumitomo-heavy-industries-Q1-2025.pdf 2. 05a_contradiction_irs-pub15_2025_SUPERSEDED.pdf 3. 06a_nearduplicate_irs-pub15b_2025.pdf 4. 04a_mixed_md_uv-changelog.md 5. 03_longform_fed-monetary-policy-report-2026-07.pdf 6. 04d_mixed_docx_newvienna-may2026.docx 7. 05b_contradiction_irs-pub15_2026_CURRENT.pdf 8. 06b_nearduplicate_irs-pub15b_2026.pdf 9. 04c_mixed_sheet_usgs-earthquakes-converted.txt 10. 04b_mixed_html_simonwillison-qwen.html 11. 07_nonenglish_hindi_rbi_mpc.pdf This is exactly the same 11-document set from the original ingestion (see 01_list_corpora.txt / 02_list_documents.txt) — no web page, URL-derived document, or crawl-originated content appears anywhere in this list. --- Verified conclusion --- The connector/pipeline run genuinely discovered and processed real pages (465 records, 100% success, per the live run screenshots/video in Input 7), but that crawled content did not land in any corpus in this account, scored or otherwise. This independently confirms the "it's unclear where the crawled content actually lands" / "zero new documents appeared in any corpus" finding in Input 7 — it is not just an assertion, it is directly checkable from this account state days after the test, with 0 new corpora and 0 new documents anywhere. --- Raw responses --- GET /v2/corpora (200): {"corpora":[{"id":"crp_3","key":"black-holes-sample-data","name":"black-holes-sample-data","description":"Information about black holes — as it turns out, they really do suck.","enabled":true,"queries_are_answers":false,"documents_are_questions":false,"encoder_id":"enc_22932","encoder_name":"boomerang-2023-q3","save_history":false,"limits":{"used_docs":1,"used_parts":470,"used_characters":142478,"max_metadata_bytes":10485760},"created_at":"2026-08-10T20:04:15.000Z"},{"id":"crp_4","key":"semiconductor_sample_data","name":"semiconductor-sample-data","description":"Information about semiconductors and their applications","enabled":true,"queries_are_answers":false,"documents_are_questions":false,"encoder_id":"enc_22932","encoder_name":"boomerang-2023-q3","save_history":false,"limits":{"used_docs":1,"used_parts":397,"used_characters":91576,"max_metadata_bytes":10485760},"created_at":"2026-08-10T20:07:28.000Z"},{"id":"crp_5","key":"rag-eval-corpus","name":"rag-eval-corpus","description":"Fixed 12-file / 7-slot eval corpus for RAG-as-a-Service platform comparison (ClickUp 86bazaep4).","enabled":true,"queries_are_answers":false,"documents_are_questions":false,"encoder_id":"enc_22932","encoder_name":"boomerang-2023-q3","save_history":false,"limits":{"used_docs":11,"used_parts":9457,"used_characters":2546762,"max_metadata_bytes":10485760},"created_at":"2026-08-18T19:54:26.000Z"},{"id":"crp_6","key":"rag-eval-corpus-api","name":"rag-eval-corpus-api","description":"API-only ingestion corpus for RAG-as-a-Service eval (ClickUp 86bazaep4) - no console UI used, per task spec (developer API, no UI).","enabled":true,"queries_are_answers":false,"documents_are_questions":false,"encoder_id":"enc_22932","encoder_name":"boomerang-2023-q3","save_history":false,"limits":{"used_docs":11,"used_parts":10072,"used_characters":2561081,"max_metadata_bytes":10485760},"created_at":"2026-08-18T21:04:43.000Z"},{"id":"crp_7","key":"rag-eval-tenant2","name":"rag-eval-tenant2","description":"Tenancy isolation test - docs 1-3 only","enabled":true,"queries_are_answers":false,"documents_are_questions":false,"encoder_id":"enc_22932","encoder_name":"boomerang-2023-q3","save_history":false,"limits":{"used_docs":2,"used_parts":1031,"used_characters":342690,"max_metadata_bytes":10485760},"created_at":"2026-08-18T21:33:35.000Z"}],"metadata":{"page_key":""}} GET /v2/corpora/rag-eval-corpus-api/documents (200, page 1 of 2, ids only for brevity — full metadata omitted here, identical to 02_list_documents.txt): ["01_tables_sumitomo-heavy-industries-Q1-2025.pdf","05a_contradiction_irs-pub15_2025_SUPERSEDED.pdf","06a_nearduplicate_irs-pub15b_2025.pdf","04a_mixed_md_uv-changelog.md","03_longform_fed-monetary-policy-report-2026-07.pdf","04d_mixed_docx_newvienna-may2026.docx","05b_contradiction_irs-pub15_2026_CURRENT.pdf","06b_nearduplicate_irs-pub15b_2026.pdf","04c_mixed_sheet_usgs-earthquakes-converted.txt","04b_mixed_html_simonwillison-qwen.html"] GET /v2/corpora/rag-eval-corpus-api/documents (200, page 2 of 2): ["07_nonenglish_hindi_rbi_mpc.pdf"]