From 194084d05864f545d62988b7b5e0b4ca57bf2272 Mon Sep 17 00:00:00 2001
From: Kailigithub <12250313+Kailigithub@users.noreply.github.com>
Date: Sat, 18 Jul 2026 03:07:58 +0800
Subject: [PATCH] fix(background-review): add relevance gate to skill review
 prompt

The per-session background review fork could write unrelated content
into a skill via skill_manage(write_file) and skill_manage(patch) when
the conversation produced research or notes that did not match the
skill's documented purpose. The 'Be ACTIVE' pressure encouraged false
positives: the fork saw research output and grafted it onto the nearest
plausible skill.

Closes #66350
---
 agent/background_review.py | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/agent/background_review.py b/agent/background_review.py
index c2ea87bd94e2..9de58092f319 100644
--- a/agent/background_review.py
+++ b/agent/background_review.py
@@ -183,6 +183,14 @@ def _digest_history(messages_snapshot: List[Dict], tail: int = 24) -> List[Dict]
     "ACTIVE — most sessions produce at least one skill update, even if "
     "small. A pass that does nothing is a missed learning opportunity, "
     "not a neutral outcome.\n\n"
+    "Relevance gate (mandatory): before any skill_manage(write_file), "
+    "skill_manage(patch), or skill_manage(create), verify that the "
+    "content is directly about the skill's documented purpose and scope. "
+    "A skill about X should not receive references about Y, even if Y is "
+    "interesting research. When in doubt, do not write — 'nothing to "
+    "save' is the correct outcome for sessions that did not involve the "
+    "skill's domain. Topical mismatch is the most common false positive; "
+    "refusing to write is a valid, valuable action.\n\n"
     "Target shape of the library: CLASS-LEVEL skills, each with a rich "
     "SKILL.md and a `references/` directory for session-specific detail. "
     "Not a long flat list of narrow one-session-one-skill entries. This "
@@ -290,8 +298,15 @@ def _digest_history(messages_snapshot: List[Dict], tail: int = 24) -> List[Dict]
     "how you should behave? Save facts about the user and durable "
     "preferences with the memory tool.\n\n"
     "**Skills**: how to do this class of task. Be ACTIVE — most "
-    "sessions produce at least one skill update. A pass that does "
-    "nothing is a missed learning opportunity, not a neutral outcome.\n\n"
+    "sessions produce at least one skill update. A pass that "
+    "does nothing is a missed learning opportunity, not a neutral outcome.\n\n"
+    "Relevance gate (mandatory): before any skill_manage(write_file), "
+    "skill_manage(patch), or skill_manage(create), verify that the "
+    "content is directly about the skill's documented purpose and scope. "
+    "A skill about X should not receive references about Y, even if Y is "
+    "interesting research. When in doubt, do not write — 'nothing to "
+    "save' is the correct outcome. Topical mismatch is the most common "
+    "false positive; refusing to write is a valid, valuable action.\n\n"
     "Target shape of the skill library: CLASS-LEVEL skills with a rich "
     "SKILL.md and a `references/` directory for session-specific detail. "
     "Not a long flat list of narrow one-session-one-skill entries.\n\n"