fix: harden reviewer workflow and memory handling

This commit is contained in:
2026-07-23 12:34:23 +02:00
parent 00761ae2ca
commit 694b7dd21f
23 changed files with 728 additions and 75 deletions
+5 -6
View File
@@ -9,8 +9,7 @@ already discarded (even after a Phase 2 reopen).
For each memory to present in the checklist, include in the option's `label` and/or
`description`:
- **What it says**: type + subject + detail (e.g. "table_promoted:
fact_seeablazione — main table for ablazioni").
- **What it says**: the clarified concept, its subject, and its full definition.
- **Where it comes from**: question_context and origin session_id.
- **Why it might apply here**: overlap of concepts/tables with the current question
(fields tables/concepts), similarity score.
@@ -19,10 +18,10 @@ For each memory to present in the checklist, include in the option's `label` and
Rules:
- Propose at most **5** candidates. Include ONLY memories of the 3 reusable types:
`concept_clarified`, `table_promoted`, `table_excluded`. Query-specific decisions
(e.g. `question_rewritten`, `sql_approved`) are NOT to be proposed: they don't
transfer to other questions.
- Propose at most **5** candidates. Include ONLY `concept_clarified` memories.
Table choices (`table_promoted`, `table_excluded`) and all other query-specific
decisions are not memories: never store, retrieve, or propose them because they
do not transfer to other questions.
- All candidate memories go in **a single** `reviewer_decide(multi:true,
advance:true, allow_empty:true)`: every option describes a candidate memory, never
an opposite action such as "do not use it". Each selected option is applied