From 95386e4a5556d63c9309bdae4c6e89d5cf655d74 Mon Sep 17 00:00:00 2001 From: Rene Fichtmueller Date: Sat, 9 May 2026 17:51:21 +0200 Subject: [PATCH] crawl: add atgbics url evidence learning --- robot-experiences/2026-05-09.jsonl | 1 + 1 file changed, 1 insertion(+) diff --git a/robot-experiences/2026-05-09.jsonl b/robot-experiences/2026-05-09.jsonl index cd98898..28f9d01 100644 --- a/robot-experiences/2026-05-09.jsonl +++ b/robot-experiences/2026-05-09.jsonl @@ -28,3 +28,4 @@ {"event":"db_evidence_backfill","observed_at":"2026-05-09T15:43:00Z","actor":"codex-fibermall-url-protocol-backfill","profile":"erik-safe-db-only","wave":"fibermall-url-protocol-closure","vendor":"FiberMall","summary":"Backfilled 12 FiberMall rows from protocol evidence in product URL slugs. Promoted 12 additional rows to fully_verified.","input":{"precheck":{"fibermall_near_complete_missing_details":36,"safe_url_protocol_rows":12}},"decision":{"rules":["Use URL protocol slugs such as LRM, LR4, 2FR4, LR8, SR4, ZR, SFP28-LR only when they directly encode form factor/speed/reach/fiber/wavelength.","Skip brand-name-only rows without protocol/reach evidence.","Correct form factor and speed when URL evidence contradicts stale parser defaults."],"runtime_policy":"DB-only update; no crawler wave; no external AI."},"outcome":{"updated":{"fibermall_detail_rows":12,"fully_verified_promoted":12},"postcheck":{"fibermall_near_complete_missing_details":24},"global_after":{"total":17647,"details_verified":11616,"fully_verified":10340},"tip_health":{"status":"healthy","load_status":"ok","memory_used_pct":13}},"truth_policy":"Remaining FiberMall rows need stronger product-page parsing; brand/OEM-code-only URL rows must not be auto-approved.","safety_notes":["No external AI was used.","No browser crawler was started.","Erik public health stayed healthy."]} {"event":"db_evidence_backfill","observed_at":"2026-05-09T15:48:00Z","actor":"codex-gbics-qsfptek-fluxlight-standard-backfill","profile":"erik-safe-db-only","wave":"small-vendor-near-complete-closure","vendor":"GBICS+QSFPTEK+Fluxlight","summary":"Backfilled 32 near-complete rows from explicit standard/title/URL evidence across GBICS, QSFPTEK and Fluxlight. All three vendor near-complete queues reached zero.","input":{"precheck":{"gbics_near_complete":13,"qsfptek_near_complete":8,"fluxlight_near_complete":11}},"decision":{"rules":["Use explicit nm/m/km evidence where present.","Use well-defined protocol codes only when paired with vendor/source URL evidence.","Correct stale parser defaults only when part number or URL encodes form factor and speed.","Do not broaden this logic to ambiguous OEM-code rows."],"runtime_policy":"DB-only update; no crawler wave; no external AI."},"outcome":{"updated":{"detail_rows":32,"fully_verified_promoted":32},"postcheck":{"gbics_near_complete":0,"qsfptek_near_complete":0,"fluxlight_near_complete":0},"global_after":{"total":17647,"details_verified":11648,"fully_verified":10372},"tip_health":{"status":"healthy","load_status":"ok","memory_used_pct":13}},"truth_policy":"This was an evidence pass, not a guess pass; only explicit standard/title/URL evidence was accepted.","safety_notes":["No external AI was used.","No browser crawler was started.","Erik public health stayed healthy."]} {"event":"db_evidence_backfill","observed_at":"2026-05-09T15:53:00Z","actor":"codex-naddod-adapter-and-fs-final-detail-closure","profile":"erik-safe-db-plus-source-search","wave":"naddod-fs-near-complete-closure","vendor":"NADDOD+FS.COM","summary":"Closed NADDOD and FS.COM near-complete detail queues: classified three NADDOD rows as adapter/converter products and corrected FS SKU 110529 to QDD-LR4-400G 400GBASE-LR4 QSFP-DD evidence.","input":{"precheck":{"naddod_near_complete":3,"fscom_near_complete":1},"source_evidence":["NADDOD product pages/products search snippets identify 100GBASE-S25, 40GBASE-S10, and MAM1Q00A-QSA28-S as QSFP/SFP adapter converter modules.","Official FS source for SKU 110529 identifies QDD-LR4-400G, 400GBASE-LR4 QSFP-DD, SMF, 10km, CWDM4 1271/1291/1311/1331nm, Duplex LC."]},"decision":{"rules":["Adapter/converter products get Adapter / Converter class and N/A optical reach/fiber/wavelength semantics.","Do not use adapter/converter rows as optical equivalence evidence.","FS SKU 110529 should be normalized to QDD-LR4-400G instead of opaque FS-110529."],"runtime_policy":"DB-only update after source lookup; no crawler wave; no external AI."},"outcome":{"updated":{"detail_rows":4,"fully_verified_promoted":3},"postcheck":{"naddod_near_complete":0,"fscom_near_complete":0},"global_after":{"total":17647,"details_verified":11652,"fully_verified":10375},"tip_health":{"status":"healthy","load_status":"ok","memory_used_pct":13}},"truth_policy":"Adapter/converter rows are product records but not optical transceiver equivalents; FS SKU 110529 is LR4, not a generic opaque product.","safety_notes":["No external AI was used.","No browser crawler was started.","FS direct page body was WAF-challenged, so official indexed FS source evidence was used instead."]} +{"event":"db_evidence_backfill","observed_at":"2026-05-09T16:00:00Z","actor":"codex-atgbics-explicit-url-evidence-backfill","profile":"erik-safe-db-only","wave":"atgbics-url-evidence-closure","vendor":"ATGBICS","summary":"Backfilled 346 ATGBICS rows from explicit product URL evidence for distance, media and wavelength. Promoted 346 additional rows to fully_verified.","input":{"precheck":{"atgbics_near_complete_missing_details":485,"safe_url_reach_media_rows":346,"url_evidence":["m/km distance","smf/mmf/copper/dac/base-t/rj45 media","nm wavelength where optical"]}},"decision":{"rules":["Only accept rows where product URL contains explicit distance and media evidence.","Extract reach from m/km URL token, not from unrelated product-number digits.","Extract wavelength from nm URL token when optical; use N/A for copper/dac/base-t/rj45.","Correct form factor and speed only from clear protocol terms in the URL."],"runtime_policy":"DB-only update; no crawler wave; no external AI."},"outcome":{"updated":{"atgbics_detail_rows":346,"fully_verified_promoted":346},"postcheck":{"atgbics_near_complete_missing_details":139},"global_after":{"total":17647,"details_verified":11998,"fully_verified":10721},"tip_health":{"status":"healthy","load_status":"ok","memory_used_pct":13}},"truth_policy":"Remaining ATGBICS rows need product-page parsing or special handling because they do not expose simple m/km plus media URL evidence.","safety_notes":["No external AI was used.","No browser crawler was started.","Erik public health stayed healthy."]}