claude-code-arcane 2.9.8 → 2.9.9

package/skills/ue-input-system/SKILL.md

@@ -383,7 +383,6 @@ Validate before considering the work done:
 - [ ] Input Action value types match how the action is consumed
 - [ ] No legacy `BindAction`/`BindAxis` or direct key polling remains
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-mass-entity/SKILL.md

@@ -465,7 +465,6 @@ Validate before considering the work done:
 - [ ] Processor execution order and dependencies are declared explicitly
 - [ ] Mass is justified by entity count; a plain actor would not do
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-materials-rendering/SKILL.md

@@ -515,7 +515,6 @@ Validate before considering the work done:
 - [ ] Instruction count and texture samplers stay within the platform budget
 - [ ] Dynamic material instances are created once and cached, not per frame
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-module-build-system/SKILL.md

@@ -476,7 +476,6 @@ Validate before considering the work done:
 - [ ] Module type and loading phase are correct for what the module does
 - [ ] The project still builds for every target platform it claims to support
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-naming-conventions/SKILL.md

@@ -130,7 +130,6 @@ Validate before considering the work done:
 - [ ] Any rename ran with clean source control and a Fix Up Redirectors pass afterwards
 - [ ] The agreed convention is written into `docs/unreal/project-context.md`
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks asset references or leaves the project inconsistent; fix it before handing the work over.
 

package/skills/ue-networking-replication/SKILL.md

@@ -440,7 +440,6 @@ Validate before considering the work done:
 - [ ] Server RPCs validate all client input; nothing trusts the client
 - [ ] Replication conditions and update frequency were set per property, not left default
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-niagara-effects/SKILL.md

@@ -520,7 +520,6 @@ Validate before considering the work done:
 - [ ] GPU simulation is used where the effect is heavy and does not need CPU readback
 - [ ] Effects are pooled or auto-destroyed rather than leaking components
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-packaging/SKILL.md

@@ -104,7 +104,6 @@ Validate before considering the work done:
 - [ ] The packaged executable was launched and verified, not just cooked successfully
 - [ ] No credentials or internal config were staged into the build
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that produces a broken or unshippable build; fix it before handing the work over.
 

package/skills/ue-physics-collision/SKILL.md

@@ -412,7 +412,6 @@ Validate before considering the work done:
 - [ ] Traces use the narrowest channel and shape that answers the question
 - [ ] Complex collision is avoided on anything queried frequently
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-procedural-generation/SKILL.md

@@ -494,7 +494,6 @@ Validate before considering the work done:
 - [ ] Instanced meshes are used for repeated geometry rather than individual actors
 - [ ] Runtime generation is budgeted and spread across frames
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-project-context/SKILL.md

@@ -293,7 +293,6 @@ Validate before considering the work done:
 - [ ] Coding conventions reflect what the codebase really does, not aspirations
 - [ ] Every other `ue-*` skill can answer its Context Check from this document
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-sequencer-cinematics/SKILL.md

@@ -495,7 +495,6 @@ Validate before considering the work done:
 - [ ] Sequences are bound through spawnables or resolvable bindings, not fragile references
 - [ ] Render settings are separated from in-game playback settings
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-serialization-savegames/SKILL.md

@@ -513,7 +513,6 @@ Validate before considering the work done:
 - [ ] Saving is asynchronous and cannot corrupt an existing file on failure
 - [ ] Only owned state is serialized — no raw object pointers to transient actors
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-state-trees/SKILL.md

@@ -478,7 +478,6 @@ Validate before considering the work done:
 - [ ] Transitions are driven by conditions and events, not by side effects in tasks
 - [ ] The schema chosen matches the owner type of the tree
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-testing-debugging/SKILL.md

@@ -508,7 +508,6 @@ Validate before considering the work done:
 - [ ] Tests run headless in CI without requiring a human in the editor
 - [ ] Logging uses a dedicated category at an appropriate verbosity
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-ui-umg-slate/SKILL.md

@@ -419,7 +419,6 @@ Validate before considering the work done:
 - [ ] Focus, navigation and gamepad input were verified, not just mouse
 - [ ] Widgets are pooled or properly removed rather than accumulating
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ue-world-level-streaming/SKILL.md

@@ -507,7 +507,6 @@ Validate before considering the work done:
 - [ ] Data layers are used for state variation instead of duplicated levels
 - [ ] Level transitions handle actors that must survive travel
 - [ ] Conventions in `docs/unreal/project-context.md` were read and followed
-- [ ] The Common Mistakes section of this skill was reviewed against the change
 
 **PASS** — every item holds. **CONCERNS** — the work stands but at least one item is unresolved and worth flagging to the user. **FAIL** — an item is violated in a way that breaks correctness, replication or the frame budget; fix it before handing the work over.
 

package/skills/ux-review/SKILL.md

@@ -65,7 +65,7 @@ Report: completeness (X/Y sections), quality issues (BLOCKING/ADVISORY with fix)
 ## Protocol
 
 - Read-only — never edits files
-- APPROVED → suggest `/team-ui`
+- APPROVED → the spec is ready for implementation
 - NEEDS REVISION → offer to help draft missing sections
 - MAJOR REVISION → redirect to `/ux-design`
 - Verdict is advisory — user decides whether to proceed

package/docs/RELEASE-SETUP.md

@@ -1,99 +0,0 @@
-# Release Setup Guide
-
-Arcane usa **semantic-release** para publicar automaticamente a npm en cada push a `main`. Esta guia explica como configurar los secrets necesarios para que el pipeline funcione.
-
-## Arquitectura del release
-
-```
-Push a main → GitHub Actions → build + test → semantic-release → npm publish + GitHub Release
-```
-
-semantic-release analiza los commits (conventional commits) para determinar la version:
-- `fix:` → patch (1.0.0 → 1.0.1)
-- `feat:` → minor (1.0.0 → 1.1.0)
-- `BREAKING CHANGE:` → major (1.0.0 → 2.0.0)
-- `chore:`, `docs:`, `test:`, `refactor:` → no release
-
-## Prerequisitos
-
-- Cuenta en [npmjs.com](https://www.npmjs.com) con acceso al paquete `claude-code-arcane`
-- Acceso de admin al repositorio en GitHub
-
-## Paso 1: Crear token en npm
-
-1. Ir a **https://www.npmjs.com** y loguearse
-2. Click en el avatar (arriba a la derecha) → **Access Tokens**
-3. Click **Generate New Token** → elegir **Granular Access Token**
-4. Configurar:
-   - **Token name:** `arcane-release`
-   - **Expiration:** sin expiracion (o la que se prefiera)
-   - **Packages and scopes:** seleccionar **Read and write**
-   - **Select packages:** elegir `claude-code-arcane`
-5. Click **Generate Token**
-6. **Copiar el token** — solo se muestra una vez
-
-## Paso 2: Agregar secret en GitHub
-
-1. Ir a [Settings > Secrets > Actions](https://github.com/SebastianLuser/Claude-Code-Arcane/settings/secrets/actions)
-2. Click **New repository secret**
-3. Completar:
-   - **Name:** `NPM_TOKEN`
-   - **Secret:** pegar el token copiado en el paso anterior
-4. Click **Add secret**
-
-> **Nota:** `GITHUB_TOKEN` se provee automaticamente por GitHub Actions, no hace falta configurarlo.
-
-## Paso 3: Verificar
-
-1. Hacer un push a `main` con un commit tipo `feat:` o `fix:`
-2. Ir a [Actions](https://github.com/SebastianLuser/Claude-Code-Arcane/actions) y verificar que el workflow **Release** corre exitosamente
-3. Verificar en npm que la nueva version aparece: `npm view claude-code-arcane version`
-4. Verificar que se creo un [GitHub Release](https://github.com/SebastianLuser/Claude-Code-Arcane/releases) con changelog automatico
-
-## Archivos de configuracion
-
-| Archivo | Proposito |
-|---------|-----------|
-| `.releaserc.json` | Configuracion de semantic-release (plugins, branches) |
-| `.github/workflows/release.yml` | Workflow de GitHub Actions que ejecuta semantic-release |
-| `.github/workflows/ci.yml` | CI separado (build + test + typecheck) |
-
-## Troubleshooting
-
-### El workflow falla con "NPM_TOKEN is not set"
-El secret no esta configurado. Seguir el Paso 2.
-
-### El workflow corre pero no publica
-semantic-release solo publica si hay commits que justifiquen un bump de version. Commits tipo `chore:`, `docs:`, `test:` no generan release. Usar `feat:` o `fix:`.
-
-### Error "npm ERR! 403 Forbidden"
-El token no tiene permisos de escritura sobre el paquete, o expiro. Regenerar siguiendo el Paso 1.
-
-### El CHANGELOG.md no se actualiza
-semantic-release commitea el changelog automaticamente con `[skip ci]` para no generar un loop. Si no aparece, verificar que el plugin `@semantic-release/git` este en `.releaserc.json`.
-
-## Hybrid Distribution
-
-Ademas del release automatico, Arcane soporta distribucion hibrida:
-
-- **GitHub source:** `npx claude-code-arcane install backend-ts --source github` baja contenido directamente del repo
-- **Cache local:** contenido descargado se cachea en `~/.arcane/cache/`
-- **Bundled fallback:** si GitHub no esta disponible, usa el contenido empaquetado en npm
-- **Auto:** `--source auto` (default) intenta GitHub → cache → bundled
-
-Variable de entorno: `ARCANE_SOURCE=bundled|github|auto`
-
-## Smart Update
-
-`arcane update` compara el contenido instalado con la ultima version disponible:
-
-```bash
-arcane update            # actualiza contenido (three-way merge)
-arcane update --dry-run  # muestra que cambiaria sin aplicar
-arcane update --force    # sobreescribe incluso archivos modificados localmente
-```
-
-El three-way merge detecta:
-- Cambios upstream que no fueron tocados localmente → **actualiza**
-- Archivos customizados localmente sin cambios upstream → **preserva**
-- Conflictos (ambos cambiaron) → **backup + actualiza + avisa**

package/docs/presentations/build_arcane_deck.py

@@ -1,310 +0,0 @@
-#!/usr/bin/env python3
-"""Genera la presentacion ejecutiva de Claude Code Arcane (8 slides, 16:9)."""
-import os
-from pptx import Presentation
-from pptx.util import Inches, Pt, Emu
-from pptx.dml.color import RGBColor
-from pptx.enum.text import PP_ALIGN, MSO_ANCHOR
-from pptx.enum.shapes import MSO_SHAPE
-from pptx.oxml.ns import qn
-
-# ---------- Tema (paleta Arcane: indigo profundo + cian) ----------
-BG       = RGBColor(0x1E, 0x1B, 0x4B)  # indigo-950 fondo
-PANEL    = RGBColor(0x2A, 0x25, 0x60)  # panel
-PANEL2   = RGBColor(0x31, 0x2E, 0x81)  # panel mas claro
-VIOLET   = RGBColor(0x5B, 0x4F, 0xE8)  # acento violeta
-CYAN     = RGBColor(0x22, 0xD3, 0xEE)  # acento cian
-WHITE    = RGBColor(0xF8, 0xFA, 0xFC)
-LIGHT    = RGBColor(0xE5, 0xE7, 0xEB)
-MUTED    = RGBColor(0x9C, 0xA3, 0xC4)
-GREEN    = RGBColor(0x34, 0xD3, 0x99)
-AMBER    = RGBColor(0xFB, 0xBF, 0x24)
-FONT     = "Calibri"
-FONT_H   = "Calibri"
-
-prs = Presentation()
-prs.slide_width  = Inches(13.333)
-prs.slide_height = Inches(7.5)
-BLANK = prs.slide_layouts[6]
-SW, SH = prs.slide_width, prs.slide_height
-
-
-def slide():
-    s = prs.slides.add_slide(BLANK)
-    r = s.shapes.add_shape(MSO_SHAPE.RECTANGLE, 0, 0, SW, SH)
-    r.fill.solid(); r.fill.fore_color.rgb = BG
-    r.line.fill.background()
-    r.shadow.inherit = False
-    return s
-
-
-def rect(s, x, y, w, h, color, line=None, radius=False, line_w=1.0):
-    shp = MSO_SHAPE.ROUNDED_RECTANGLE if radius else MSO_SHAPE.RECTANGLE
-    sp = s.shapes.add_shape(shp, Inches(x), Inches(y), Inches(w), Inches(h))
-    if color is None:
-        sp.fill.background()
-    else:
-        sp.fill.solid(); sp.fill.fore_color.rgb = color
-    if line is None:
-        sp.line.fill.background()
-    else:
-        sp.line.color.rgb = line; sp.line.width = Pt(line_w)
-    sp.shadow.inherit = False
-    return sp
-
-
-def txt(s, x, y, w, h, runs, align=PP_ALIGN.LEFT, anchor=MSO_ANCHOR.TOP,
-        space_after=4, line_spacing=1.0):
-    """runs: lista de parrafos; cada parrafo es lista de (texto, size, color, bold)."""
-    tb = s.shapes.add_textbox(Inches(x), Inches(y), Inches(w), Inches(h))
-    tf = tb.text_frame; tf.word_wrap = True
-    tf.vertical_anchor = anchor
-    for i, para in enumerate(runs):
-        p = tf.paragraphs[0] if i == 0 else tf.add_paragraph()
-        p.alignment = align
-        p.space_after = Pt(space_after); p.space_before = Pt(0)
-        p.line_spacing = line_spacing
-        for (t, sz, col, bold) in para:
-            r = p.add_run(); r.text = t
-            r.font.size = Pt(sz); r.font.color.rgb = col
-            r.font.bold = bold; r.font.name = FONT
-    return tb
-
-
-def badge(s, n):
-    c = s.shapes.add_shape(MSO_SHAPE.OVAL, Inches(12.55), Inches(6.85), Inches(0.42), Inches(0.42))
-    c.fill.solid(); c.fill.fore_color.rgb = VIOLET; c.line.fill.background(); c.shadow.inherit = False
-    tf = c.text_frame; tf.word_wrap = False
-    p = tf.paragraphs[0]; p.alignment = PP_ALIGN.CENTER
-    r = p.add_run(); r.text = str(n)
-    r.font.size = Pt(12); r.font.bold = True; r.font.color.rgb = WHITE; r.font.name = FONT
-    tf.vertical_anchor = MSO_ANCHOR.MIDDLE
-
-
-def footer(s):
-    txt(s, 0.55, 7.0, 5, 0.35, [[("Claude Code Arcane", 9, MUTED, False)]])
-
-
-def kicker(s, text, color=CYAN):
-    rect(s, 0.55, 0.55, 0.12, 0.55, color)
-    txt(s, 0.8, 0.5, 11, 0.5, [[(text, 13, color, True)]])
-
-
-def title(s, text, y=0.95, size=34):
-    txt(s, 0.78, y, 11.8, 1.0, [[(text, size, WHITE, True)]])
-
-
-# ================= SLIDE 1 — Portada =================
-s = slide()
-rect(s, 0, 0, 13.333, 7.5, None)
-# barra de acento lateral
-rect(s, 0, 0, 0.25, 7.5, VIOLET)
-rect(s, 0.25, 0, 0.08, 7.5, CYAN)
-txt(s, 1.0, 1.7, 11.5, 0.6, [[("CLAUDE CODE", 20, CYAN, True)]])
-txt(s, 1.0, 2.15, 11.5, 1.4, [[("Arcane", 88, WHITE, True)]])
-txt(s, 1.05, 3.65, 11, 0.6,
-    [[("Un gestor de paquetes de configuracion para Claude Code", 22, LIGHT, False)]])
-txt(s, 1.05, 4.25, 11, 0.5,
-    [[("Instala solo las capacidades que tu proyecto necesita — con un comando.", 15, MUTED, False)]])
-
-# dashboard de metricas
-metrics = [("322", "Skills"), ("86", "Agentes"), ("29", "Profiles"), ("19", "Rules")]
-mx, mw, gap = 1.0, 2.7, 0.25
-for i, (num, lbl) in enumerate(metrics):
-    x = mx + i * (mw + gap)
-    rect(s, x, 5.3, mw, 1.35, PANEL, radius=True)
-    rect(s, x, 5.3, 0.1, 1.35, CYAN if i % 2 else VIOLET, radius=False)
-    txt(s, x, 5.45, mw, 0.8, [[(num, 44, WHITE, True)]], align=PP_ALIGN.CENTER)
-    txt(s, x, 6.2, mw, 0.4, [[(lbl.upper(), 13, CYAN, True)]], align=PP_ALIGN.CENTER)
-
-# ================= SLIDE 2 — El problema =================
-s = slide()
-kicker(s, "POR QUE EXISTE")
-title(s, "El problema que resuelve")
-problems = [
-    ("Configuracion manual", "Copiar .claude/ a mano en cada proyecto: repetitivo y diverge con el tiempo."),
-    ("Todo-o-nada", "Un mega-repo cargaria las 322 skills en CADA proyecto: mas tokens, mas ruido, mas latencia."),
-    ("Sin estandarizacion", "Cada dev configura su Claude distinto. No hay fuente de verdad compartida."),
-    ("Sin actualizacion", "Una mejora en una skill nunca llega a los proyectos que ya la copiaron."),
-]
-y = 2.05
-for i, (h, d) in enumerate(problems):
-    rect(s, 0.78, y, 11.7, 1.0, PANEL, radius=True)
-    rect(s, 0.78, y, 0.1, 1.0, VIOLET)
-    txt(s, 1.05, y + 0.13, 3.4, 0.8, [[(h, 17, CYAN, True)]], anchor=MSO_ANCHOR.MIDDLE)
-    txt(s, 4.5, y + 0.13, 7.8, 0.8, [[(d, 13.5, LIGHT, False)]], anchor=MSO_ANCHOR.MIDDLE)
-    y += 1.15
-txt(s, 0.78, 6.75, 11.8, 0.5,
-    [[("Falta una forma modular, versionada y compartible de configurar Claude Code.", 15, WHITE, True)]])
-badge(s, 2); footer(s)
-
-# ================= SLIDE 3 — Idea central =================
-s = slide()
-kicker(s, "LA IDEA CENTRAL")
-title(s, "npm, pero para las capacidades de Claude Code")
-txt(s, 0.78, 2.0, 11.7, 1.0,
-    [[("En vez de instalar librerias de codigo, instalas ", 17, LIGHT, False),
-      ("capacidades", 17, CYAN, True),
-      (": hacer commits, disenar APIs, correr un sprint, auditar seguridad — adaptadas a tu stack.", 17, LIGHT, False)]],
-    line_spacing=1.15)
-
-# comando hero estilo terminal
-rect(s, 0.78, 3.25, 11.7, 1.5, RGBColor(0x14, 0x11, 0x33), radius=True, line=VIOLET, line_w=1.5)
-txt(s, 1.1, 3.45, 11, 0.4, [[("terminal", 11, MUTED, False)]])
-txt(s, 1.1, 3.85, 11, 0.7,
-    [[("$ ", 26, CYAN, True), ("npx claude-code-arcane install ", 26, WHITE, True),
-      ("backend-ts", 26, GREEN, True), ("+agile", 26, AMBER, True), ("+testing", 26, AMBER, True)]])
-
-cols = [("backend-ts", "base", GREEN), ("+agile", "addon", AMBER), ("+testing", "addon", AMBER)]
-cx = 0.78
-for name, kind, col in cols:
-    rect(s, cx, 5.05, 3.0, 1.1, PANEL, radius=True)
-    txt(s, cx, 5.2, 3.0, 0.5, [[(name, 18, col, True)]], align=PP_ALIGN.CENTER)
-    txt(s, cx, 5.7, 3.0, 0.4, [[(kind.upper(), 12, MUTED, True)]], align=PP_ALIGN.CENTER)
-    cx += 3.25
-txt(s, 0.78, 6.45, 11.7, 0.5,
-    [[("base + addon + addon  ->  solo lo que ESE proyecto necesita.", 15, WHITE, True)]])
-badge(s, 3); footer(s)
-
-# ================= SLIDE 4 — Como funciona =================
-s = slide()
-kicker(s, "ARQUITECTURA")
-title(s, "Como funciona")
-steps = [
-    ("1", "Resuelve el contenido", "GitHub -> cache local -> bundled (funciona offline)"),
-    ("2", "Carga core.yaml siempre", "21 skills base + 15 hooks + permisos de seguridad"),
-    ("3", "Mergea los profiles", "Combina los profiles pedidos con +"),
-    ("4", "Deduplica", "Un skill repetido se copia una sola vez"),
-    ("5", "Genera .claude/", "settings.json + manifest + skills, agents, rules, hooks"),
-]
-y = 2.05
-for num, h, d in steps:
-    rect(s, 0.78, y, 11.7, 0.84, PANEL, radius=True)
-    c = s.shapes.add_shape(MSO_SHAPE.OVAL, Inches(0.98), Inches(y + 0.17), Inches(0.5), Inches(0.5))
-    c.fill.solid(); c.fill.fore_color.rgb = VIOLET; c.line.fill.background(); c.shadow.inherit = False
-    ctf = c.text_frame; ctf.vertical_anchor = MSO_ANCHOR.MIDDLE
-    cp = ctf.paragraphs[0]; cp.alignment = PP_ALIGN.CENTER
-    cr = cp.add_run(); cr.text = num; cr.font.size = Pt(20); cr.font.bold = True
-    cr.font.color.rgb = WHITE; cr.font.name = FONT
-    txt(s, 1.75, y + 0.05, 4.0, 0.74, [[(h, 16, CYAN, True)]], anchor=MSO_ANCHOR.MIDDLE)
-    txt(s, 5.9, y + 0.05, 6.4, 0.74, [[(d, 13, LIGHT, False)]], anchor=MSO_ANCHOR.MIDDLE)
-    y += 0.95
-badge(s, 4); footer(s)
-
-# ================= SLIDE 5 — 4 piezas =================
-s = slide()
-kicker(s, "SEPARACION DE RESPONSABILIDADES")
-title(s, "Cuatro piezas, cuatro responsabilidades")
-cards = [
-    ("Skills", "Acciones que Claude ejecuta", "/commit  ·  /code-review", CYAN),
-    ("Agents", "Personas que Claude adopta", "backend-architect  ·  game-designer", VIOLET),
-    ("Hooks", "Automatizacion invisible (eventos)", "valida commits  ·  escanea secrets", GREEN),
-    ("Rules", "Estandares que Claude obedece", "\"Clean architecture en backend\"", AMBER),
-]
-positions = [(0.78, 2.05), (6.93, 2.05), (0.78, 4.25), (6.93, 4.25)]
-cw, ch = 5.6, 2.0
-for (name, what, ex, col), (x, y) in zip(cards, positions):
-    rect(s, x, y, cw, ch, PANEL, radius=True)
-    rect(s, x, y, cw, 0.12, col)
-    txt(s, x + 0.3, y + 0.25, cw - 0.6, 0.5, [[(name, 24, col, True)]])
-    txt(s, x + 0.3, y + 0.85, cw - 0.6, 0.5, [[(what, 14.5, LIGHT, False)]])
-    txt(s, x + 0.3, y + 1.4, cw - 0.6, 0.5, [[(ex, 13, MUTED, True)]])
-txt(s, 0.78, 6.55, 11.7, 0.5,
-    [[("Todo en Markdown + YAML: legible, diffeable en git, editable por cualquiera.", 14, WHITE, True)]])
-badge(s, 5); footer(s)
-
-# ================= SLIDE 6 — Por que gana =================
-s = slide()
-kicker(s, "DIFERENCIADORES")
-title(s, "Por que Arcane gana")
-# encabezados
-rect(s, 0.78, 2.0, 5.7, 0.55, PANEL2, radius=True)
-rect(s, 6.63, 2.0, 5.85, 0.55, PANEL2, radius=True)
-txt(s, 1.0, 2.05, 5.3, 0.45, [[("ALTERNATIVA", 13, MUTED, True)]], anchor=MSO_ANCHOR.MIDDLE)
-txt(s, 6.85, 2.05, 5.4, 0.45, [[("COMO GANA ARCANE", 13, CYAN, True)]], anchor=MSO_ANCHOR.MIDDLE)
-rows = [
-    ("Config a mano", "Un comando, versionado, reproducible"),
-    ("Mega-repo monolitico", "Selectivo por profile + dedup (menos tokens)"),
-    ("Copiar config entre proyectos", "Smart update con three-way merge (no pisa tus cambios)"),
-    ("Claude \"vanilla\"", "Hooks + rules + 86 agentes especializados"),
-]
-y = 2.65
-for a, b in rows:
-    rect(s, 0.78, y, 5.7, 0.7, PANEL, radius=True)
-    rect(s, 6.63, y, 5.85, 0.7, PANEL, radius=True)
-    txt(s, 1.0, y + 0.05, 5.3, 0.6, [[(a, 14, LIGHT, False)]], anchor=MSO_ANCHOR.MIDDLE)
-    txt(s, 6.85, y + 0.05, 5.4, 0.6, [[(b, 13.5, WHITE, True)]], anchor=MSO_ANCHOR.MIDDLE)
-    y += 0.8
-rect(s, 0.78, y + 0.05, 11.7, 0.7, VIOLET, radius=True)
-txt(s, 0.78, y + 0.1, 11.7, 0.6,
-    [[("Trata la configuracion de IA como software de primera clase: versionado, modular, testeable y distribuido.",
-       14, WHITE, True)]], align=PP_ALIGN.CENTER, anchor=MSO_ANCHOR.MIDDLE)
-badge(s, 6); footer(s)
-
-# ================= SLIDE 7 — Pros y contras =================
-s = slide()
-kicker(s, "EVALUACION HONESTA")
-title(s, "Trade-offs")
-# Fortalezas
-rect(s, 0.78, 2.0, 5.7, 4.6, PANEL, radius=True)
-rect(s, 0.78, 2.0, 5.7, 0.6, RGBColor(0x14, 0x3D, 0x2E), radius=True)
-txt(s, 1.05, 2.05, 5.3, 0.5, [[("FORTALEZAS", 16, GREEN, True)]], anchor=MSO_ANCHOR.MIDDLE)
-pros = [
-    "Eficiencia de tokens (solo lo relevante)",
-    "Un comando, cero friccion (npx)",
-    "Reproducible (manifest = snapshot exacto)",
-    "Estandarizacion de equipo",
-    "Auto-actualizable sin pisar tus cambios",
-    "Testeable (framework skills-selftest)",
-]
-y = 2.85
-for p in pros:
-    txt(s, 1.05, y, 5.2, 0.5, [[("+  ", 14, GREEN, True), (p, 13.5, LIGHT, False)]])
-    y += 0.6
-# Costos
-rect(s, 6.93, 2.0, 5.55, 4.6, PANEL, radius=True)
-rect(s, 6.93, 2.0, 5.55, 0.6, RGBColor(0x4A, 0x39, 0x12), radius=True)
-txt(s, 7.2, 2.05, 5.1, 0.5, [[("COSTOS / TRADE-OFFS", 16, AMBER, True)]], anchor=MSO_ANCHOR.MIDDLE)
-cons = [
-    "Mantener 322 skills con calidad consistente",
-    "Skills son prompts -> no deterministas",
-    "Dependencia de la API de Claude Code",
-    "Curva de aprendizaje para combinar profiles",
-]
-y = 2.95
-for c in cons:
-    txt(s, 7.2, y, 5.0, 0.6, [[("!  ", 14, AMBER, True), (c, 13.5, LIGHT, False)]])
-    y += 0.75
-badge(s, 7); footer(s)
-
-# ================= SLIDE 8 — Cierre =================
-s = slide()
-rect(s, 0, 0, 0.25, 7.5, VIOLET)
-rect(s, 0.25, 0, 0.08, 7.5, CYAN)
-kicker(s, "PRODUCTION-GRADE")
-title(s, "No es un script: es infraestructura", size=32)
-feats = [
-    ("Distribucion hibrida", "GitHub -> cache -> bundled. Funciona incluso offline."),
-    ("Smart update", "Three-way merge: actualiza sin pisar tus customizaciones."),
-    ("Worktree-aware", "Comparte hooks/docs, copia skills por rama."),
-    ("Auto-release", "semantic-release: conventional commits -> publish a npm."),
-]
-positions = [(0.78, 2.15), (6.93, 2.15), (0.78, 3.75), (6.93, 3.75)]
-for (h, d), (x, y) in zip(feats, positions):
-    rect(s, x, y, 5.6, 1.4, PANEL, radius=True)
-    rect(s, x, y, 0.1, 1.4, CYAN)
-    txt(s, x + 0.3, y + 0.18, 5.1, 0.5, [[(h, 17, CYAN, True)]])
-    txt(s, x + 0.3, y + 0.68, 5.1, 0.6, [[(d, 13, LIGHT, False)]])
-
-rect(s, 0.78, 5.55, 11.7, 1.3, VIOLET, radius=True)
-txt(s, 0.78, 5.7, 11.7, 0.55,
-    [[("Configuracion de IA, modular y versionada — un comando para empezar.", 19, WHITE, True)]],
-    align=PP_ALIGN.CENTER)
-txt(s, 0.78, 6.25, 11.7, 0.5,
-    [[("npx claude-code-arcane install <profile>", 22, WHITE, True)]], align=PP_ALIGN.CENTER)
-badge(s, 8); footer(s)
-
-out = os.path.join(os.path.dirname(os.path.abspath(__file__)), "arcane-overview.pptx")
-prs.save(out)
-print("OK ->", out)
-print("Slides:", len(prs.slides._sldIdLst))