diff --git a/content/posts/commits/2026-07-08-commit-af4788a.md b/content/posts/commits/2026-07-08-commit-af4788a.md
new file mode 100644
index 0000000..2610848
--- /dev/null
+++ b/content/posts/commits/2026-07-08-commit-af4788a.md
@@ -0,0 +1,46 @@
+---
+title: "[bojemoi] fix(grafana,ml-threat): structured RETRAIN_ logs + dashboard aligned on log format"
+date: 2026-07-08T13:20:40+02:00
+draft: false
+tags: ["commit", "bojemoi", "main"]
+categories: ["Git Activity"]
+summary: "Commit af4788a par grafana-watcher dans bojemoi"
+author: "grafana-watcher"
+---
+
+## Commit `af4788a`
+
+| | |
+|---|---|
+| **Repository** | bojemoi |
+| **Branch** | `main` |
+| **Author** | grafana-watcher |
+| **Hash** | `af4788ac84d154cee792a0172408bd241608077d` |
+
+
+### Description
+
+- api.py: emit RETRAIN_METRICS, RETRAIN_CLASS {class}, RETRAIN_DIST lines
+  after each retrain for Loki/Grafana parsing
+- dashboard: update all LogQL queries to use |= "RETRAIN_*" prefix filters
+  instead of fragile class-name + keyword combinations
+- Panel 5 renamed from "Pwned Hosts" to "Malicious Test Samples" (support
+  count from RETRAIN_CLASS malicious)
+- Panel 9 log stream now shows all RETRAIN_ lines
+
+Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
+
+### Files Changed
+
+```
+M	ml-threat/api.py
+M	volumes/grafana/dashboards/security/ml-threat-quality.json
+```
+
+### Diff Summary
+
+```
+ ml-threat/api.py                                   |  34 ++++-
+ .../dashboards/security/ml-threat-quality.json     | 155 +++++++++++++++------
+ 2 files changed, 143 insertions(+), 46 deletions(-)
+```