style: place INSTALLATION.md Windows Notes where its ToC entry says; PEP 8 blank lines in encoding-fix scripts

Review follow-up on PR #972: the Windows Notes section now sits between
Verification & Testing and Troubleshooting, matching the Table of Contents
order, and the four scripts patched for #969 are back to exactly two blank
lines after the reconfigure block.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Qgc6RYXWJPr5oW9DHU7zR4

INSTALLATION.md

@@ -21,39 +21,6 @@ Complete installation guide for all 205+ production-ready skills across multiple
 
 ---
 
-## Windows Notes
-
-Two Windows-specific caveats to know before cloning or running skills (issues #968 and #969):
-
-### Symlinks: mirror trees need `core.symlinks=true`
-
-The cross-platform mirror trees (`.gemini/`, `.codex/`, `.vibe/`, `.hermes/`) are built from **relative symlinks** into the real skill directories. Git for Windows defaults to `core.symlinks=false`, so a default clone checks those 1,300+ links out as **1-line text files containing only the target path** — copying a skill folder from a mirror tree then silently gives you dead pointer files instead of skills.
-
-Before cloning on Windows:
-
-1. Enable **Developer Mode** (Settings → System → For developers), or run Git from an elevated prompt.
-2. Clone with symlinks enabled:
-
-   ```powershell
-   git clone -c core.symlinks=true https://github.com/alirezarezvani/claude-skills.git
-   ```
-
-If you already cloned without it, either re-clone as above, or **copy skills from the real domain directories** (e.g. `engineering-team/`, `marketing-skill/`) instead of the mirror trees — the Claude Code plugin/marketplace path always uses the real directories and is unaffected. Quick sanity check after any copy: a real `SKILL.md` is more than one line long.
-
-### Console encoding: run Python tools with UTF-8
-
-Windows consoles often default to a legacy codepage (e.g. cp1252) that cannot encode the box-drawing and comparison characters (`╔`, `≥`, `✅`) some skill tools print, which crashes the script at print time with `UnicodeEncodeError`. The most-affected tools now re-encode their own output, but as a blanket fix for every script in this library, enable Python's UTF-8 mode:
-
-```powershell
-# Per session
-$env:PYTHONUTF8 = "1"
-
-# Or permanently
-setx PYTHONUTF8 1
-```
-
----
-
 ## Quick Start
 
 **Choose your agent:**
@@ -550,6 +517,39 @@ python3 ~/.claude/skills/content-production/scripts/seo_optimizer.py test-articl
 
 ---
 
+## Windows Notes
+
+Two Windows-specific caveats to know before cloning or running skills (issues #968 and #969):
+
+### Symlinks: mirror trees need `core.symlinks=true`
+
+The cross-platform mirror trees (`.gemini/`, `.codex/`, `.vibe/`, `.hermes/`) are built from **relative symlinks** into the real skill directories. Git for Windows defaults to `core.symlinks=false`, so a default clone checks those 1,300+ links out as **1-line text files containing only the target path** — copying a skill folder from a mirror tree then silently gives you dead pointer files instead of skills.
+
+Before cloning on Windows:
+
+1. Enable **Developer Mode** (Settings → System → For developers), or run Git from an elevated prompt.
+2. Clone with symlinks enabled:
+
+   ```powershell
+   git clone -c core.symlinks=true https://github.com/alirezarezvani/claude-skills.git
+   ```
+
+If you already cloned without it, either re-clone as above, or **copy skills from the real domain directories** (e.g. `engineering-team/`, `marketing-skill/`) instead of the mirror trees — the Claude Code plugin/marketplace path always uses the real directories and is unaffected. Quick sanity check after any copy: a real `SKILL.md` is more than one line long.
+
+### Console encoding: run Python tools with UTF-8
+
+Windows consoles often default to a legacy codepage (e.g. cp1252) that cannot encode the box-drawing and comparison characters (`╔`, `≥`, `✅`) some skill tools print, which crashes the script at print time with `UnicodeEncodeError`. The most-affected tools now re-encode their own output, but as a blanket fix for every script in this library, enable Python's UTF-8 mode:
+
+```powershell
+# Per session
+$env:PYTHONUTF8 = "1"
+
+# Or permanently
+setx PYTHONUTF8 1
+```
+
+---
+
 ## Troubleshooting
 
 ### Universal Installer Issues

engineering/skills/skill-security-auditor/scripts/skill_security_auditor.py

@@ -36,7 +36,6 @@ if hasattr(sys.stderr, "reconfigure"):
     sys.stderr.reconfigure(encoding="utf-8", errors="replace")
 
 
-
 class Severity(IntEnum):
     INFO = 0
     HIGH = 1

research/dossier/skills/dossier/scripts/citation_tracker.py

@@ -40,7 +40,6 @@ if hasattr(sys.stderr, "reconfigure"):
     sys.stderr.reconfigure(encoding="utf-8", errors="replace")
 
 
-
 SESSIONS_DIR = Path.home() / ".dossier_sessions"
 VALID_CLASSIFICATIONS = ["supporting", "disconfirming", "inconclusive"]
 VALID_TIERS = ["primary", "secondary", "tertiary"]

research/dossier/skills/dossier/scripts/disconfirming_evidence_balance.py

@@ -35,7 +35,6 @@ if hasattr(sys.stderr, "reconfigure"):
     sys.stderr.reconfigure(encoding="utf-8", errors="replace")
 
 
-
 SESSIONS_DIR = Path.home() / ".dossier_sessions"
 MIN_RATIO = 0.30
 WARN_RATIO = 0.20

research/dossier/skills/dossier/scripts/source_tier_classifier.py

@@ -37,7 +37,6 @@ if hasattr(sys.stderr, "reconfigure"):
     sys.stderr.reconfigure(encoding="utf-8", errors="replace")
 
 
-
 # Pattern-based tier assignment. Most specific patterns first.
 
 PRIMARY_DOMAIN_EXACT = {