diff --git a/.github/workflows/context-snapshot.yml b/.github/workflows/context-snapshot.yml
index 9183653..1810c2f 100644
--- a/.github/workflows/context-snapshot.yml
+++ b/.github/workflows/context-snapshot.yml
@@ -1,4 +1,3 @@
-
 name: Context Snapshot (mARCHCode)
 
 on:
@@ -22,15 +21,19 @@ jobs:
           python-version: "3.11"
 
       - name: Install deps
-        run: pip install pyyaml
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install pyyaml
 
       - name: Create dirs
-        run: mkdir -p scripts .archcode
+        run: |
+          mkdir -p scripts .archcode
 
       - name: Write context_snapshot.py
         shell: bash
         run: |
           cat > scripts/context_snapshot.py <<'PY'
+          #!/usr/bin/env python3
           from __future__ import annotations
           """
           Minimal context snapshot (stable)
@@ -298,7 +301,7 @@ jobs:
       - name: Generate HTML page + true TiddlyWiki .tid (no truncation)
         shell: bash
         run: |
-          python - <<'PY'
+          cat > .archcode/_context_snapshot_html_builder.py <<'PY'
           import yaml, html, sys, json
           from pathlib import Path
           from datetime import datetime
@@ -332,11 +335,9 @@ jobs:
           </style>
           """
 
-          # ---- build the HTML *fragment* (for TiddlyWiki) ----
           frag = []
           frag.append(f"<h1>Context Snapshot — {proj or '—'}</h1>")
 
-          # metadata
           frag.append("<div class='box'><h2>Métadonnées</h2><ul class='meta'>")
           for k in ("root","generated_at","python","platform","files_count","py_files_count","ignored_dirs"):
             v = meta.get(k, "")
@@ -344,12 +345,10 @@ jobs:
             frag.append(f"<li><strong>{html.escape(str(k))}</strong>: {html.escape(str(v))}</li>")
           frag.append("</ul></div>")
 
-          # tree
           frag.append("<div class='box'><h2>Arbre du projet</h2><pre>")
           frag.append(html.escape(s.get("tree","") or ""))
           frag.append("</pre></div>")
 
-          # files
           files = s.get("files") or []
           frag.append(f"<div class='box'><h2>Fichiers ({len(files)})</h2>")
           for f in files:
@@ -371,7 +370,7 @@ jobs:
 
             if defs:
               frag.append(f"<div><strong>Defs ({ndefs})</strong><ul>")
-              for d in defs:  # pas de limite: on affiche tout
+              for d in defs:
                 q = d.get("qualname","")
                 ln = d.get("lineno","")
                 params = d.get("params") or []
@@ -395,29 +394,24 @@ jobs:
                 frag.append(li)
               frag.append("</ul></div>")
             frag.append("</details>")
-          frag.append("</div>")  # end files
+          frag.append("</div>")
 
           fragment_html = CSS + "\n" + "\n".join(frag)
 
-          # ---- write standalone HTML page ----
           full_html = "<!doctype html>\n<html><head><meta charset='utf-8'><title>Context Snapshot — %s</title></head><body>%s</body></html>" % (proj, fragment_html)
           out_html.write_text(full_html, encoding="utf-8")
 
-          # ---- write true TiddlyWiki .tid ----
-          # TiddlyWiki single-tiddler format: header fields, blank line, then content.
-          # We store the *fragment* only (no <html> wrapper), as 'text/html' type.
           def tw_timestamp(dt):
-              # YYYYMMDDhhmmssmmm (UTC)
               return dt.strftime("%Y%m%d%H%M%S") + f"{dt.microsecond//1000:03d}"
 
           now = datetime.utcnow()
           title = f"Context Snapshot — {meta.get('project') or '-'}"
           header = [
-            f"title: {title}",
-            "tags: mARCHCode Snapshot",
-            f"created: {tw_timestamp(now)}",
-            f"modified: {tw_timestamp(now)}",
-            "type: text/html",
+              f"title: {title}",
+              "tags: mARCHCode Snapshot",
+              f"created: {tw_timestamp(now)}",
+              f"modified: {tw_timestamp(now)}",
+              "type: text/html",
           ]
           out_tid.write_text("\n".join(header) + "\n\n" + fragment_html, encoding="utf-8")
 
