Bootstrap architecture rules from task history

README.md

@@ -57,6 +57,11 @@ Optional tuning:
 - `MODELHUB_QUEUE_CLEANUP_REPORT_PATH` default `.modelhub_state/queue_cleanup_latest.json`
 - `MODELHUB_ARCHITECTURE_BLACKLIST_PATH` default `.modelhub_state/architecture_compatibility_blacklist.json`
 - `MODELHUB_ARCHITECTURE_BLOCK_TTL_DAYS` default `30`
+- `MODELHUB_ARCHITECTURE_COMMUNITY_PROBE_SIZE` default `50`
+- `MODELHUB_ARCHITECTURE_COMMUNITY_LOOKBACK_DAYS` default `30`
+- `MODELHUB_ARCHITECTURE_COMMUNITY_LATEST_LIMIT` default `5000`
+- `MODELHUB_ARCHITECTURE_BOOTSTRAP_WORKERS` default `8`
+- `MODELHUB_ARCHITECTURE_BOOTSTRAP_MAX_LOGS` default `0` (unlimited)
 - `MODELHUB_RECENT_MODEL_RESERVE_SLOTS` default `10` per account
 - `MODELHUB_DYNAMIC_OLD_MODEL_CLEANUP_RESERVE_SLOTS` default `5` per account
 - `MODELHUB_RECENT_MODEL_DAYS` default `7`
@@ -126,6 +131,18 @@ change that window. The stats report exposes `architectureCompatibilityBlocks`
 and per-GPU/framework block counts. The live snapshot is written to
 `.modelhub_state/architecture_compatibility_blacklist.json`.
 
+Before the first cleanup/submission cycle, the poller probes the latest public
+failed-validation records for usable failure archives. If the public endpoint
+does not expose those archives (the current API behavior), it falls back to a
+parallel, complete history scan of every configured account. Ledger metadata is
+used when present; otherwise the task type is recovered from ModelHub's task
+level and the selected framework is conservatively recovered from the target
+container image in the failure archive. All usable historical failures are
+classified once at startup and cached in `outcomes/submissions.jsonl`; subsequent
+poll cycles continue incremental learning every three cycles. Empty or
+incomplete public evidence is logged as a fallback, never as proof that no
+architecture incompatibility exists.
+
 Before a candidate reaches the submit queue, failure-informed preflight checks
 the actual ModelScope repository structure and file sizes. Non-GGUF text
 frameworks require root-level config, weights, and tokenizer assets. The memory
@@ -286,12 +303,16 @@ evidence clear stale blocks.
 Version `2026.08.12.3` extracts unsupported `model_type`/`architectures` from the
 platform's fixed failure wording, persists a dynamically growing blacklist, and
 immediately removes exact-matching waiting tasks with two active-state checks.
+Version `2026.08.12.4` bootstraps architecture feedback before the first cleanup:
+it prefers recent public failure details when available, otherwise scans the
+complete history of every configured account, recovers missing task/framework
+metadata from task levels and target images, then continues incremental learning.
 
 ## Deploy
 
 Create a tag and submit the repository URL plus tag in "我的适配智能体".
 
 ```bash
-git tag agent-v21
-git push origin agent-v21
+git tag agent-v22
+git push origin agent-v22
 ```