Feature/lmstudio provider (#30)

* chore: just a plan update

* Add LM Studio as local AI provider (OpenAI-compatible, like Ollama)

* Convert WebP thumbnails to JPEG before image analysis for LM Studio compatibility

* Strengthen language enforcement in image analysis prompt for local models

* Use i18n localized prompts for image analysis instead of English instructions

* Add airplane mode (Flugmodus) with status bar toggle and offline model preferences

* Fix flightmode: persist model IDs, skip network when offline, airplane icon

* Auto-fallback to offline models in airplane mode for chat, title, and image analysis

* Auto-select first local model as offline fallback when no explicit offline model configured

* Block git fetch/pull/push and site upload in airplane mode

* fix: thumbnails optimized for AI

* fix: error handling in airplane mode

---------

Co-authored-by: hugo <hugoms@me.com>
This commit is contained in:
Georg Bauer
2026-03-02 13:35:42 +01:00
committed by GitHub
parent 4b4a9c1c8b
commit 5747925503
34 changed files with 2215 additions and 105 deletions

View File

@@ -78,5 +78,7 @@
"render.month.9": "Sept.",
"render.month.10": "Oktober",
"render.month.11": "Nov.",
"render.month.12": "Dezember"
"render.month.12": "Dezember",
"ai.imageAnalysis.system": "Du erzeugst Bild-Metadaten. Schreibe alle Werte auf Deutsch.\n\nRegeln:\n- \"title\": kurzer beschreibender Titel (3-8 Wörter)\n- \"alt\": sachliche Beschreibung des Sichtbaren (5-12 Wörter). Keine Interpretationen. Kein Präfix \"Bild von\".\n- \"caption\": ansprechende Blog-Bildunterschrift (5-20 Wörter)\n\nAntworte ausschließlich mit JSON: {\"title\": \"...\", \"alt\": \"...\", \"caption\": \"...\"}",
"ai.imageAnalysis.user": "Analysiere dieses Bild. Antworte mit JSON auf Deutsch."
}

View File

@@ -78,5 +78,7 @@
"render.month.9": "September",
"render.month.10": "October",
"render.month.11": "November",
"render.month.12": "December"
"render.month.12": "December",
"ai.imageAnalysis.system": "You generate image metadata. Write all values in English.\n\nRules:\n- \"title\": short descriptive title (3-8 words)\n- \"alt\": factual description of what is visible (5-12 words). No interpretations. No \"Image of\" prefix.\n- \"caption\": engaging blog caption (5-20 words)\n\nRespond with JSON only: {\"title\": \"...\", \"alt\": \"...\", \"caption\": \"...\"}",
"ai.imageAnalysis.user": "Analyze this image. Respond with JSON in English."
}

View File

@@ -78,5 +78,7 @@
"render.month.9": "septiembre",
"render.month.10": "octubre",
"render.month.11": "noviembre",
"render.month.12": "diciembre"
"render.month.12": "diciembre",
"ai.imageAnalysis.system": "Generas metadatos de imagen. Escribe todos los valores en español.\n\nReglas:\n- \"title\": título descriptivo corto (3-8 palabras)\n- \"alt\": descripción factual de lo visible (5-12 palabras). Sin interpretaciones. Sin prefijo \"Imagen de\".\n- \"caption\": pie de foto atractivo para blog (5-20 palabras)\n\nResponde solo con JSON: {\"title\": \"...\", \"alt\": \"...\", \"caption\": \"...\"}",
"ai.imageAnalysis.user": "Analiza esta imagen. Responde con JSON en español."
}

View File

@@ -78,5 +78,7 @@
"render.month.9": "septembre",
"render.month.10": "octobre",
"render.month.11": "novembre",
"render.month.12": "décembre"
"render.month.12": "décembre",
"ai.imageAnalysis.system": "Tu génères des métadonnées d'image. Écris toutes les valeurs en français.\n\nRègles :\n- \"title\" : titre descriptif court (3-8 mots)\n- \"alt\" : description factuelle de ce qui est visible (5-12 mots). Pas d'interprétations. Pas de préfixe \"Image de\".\n- \"caption\" : légende de blog engageante (5-20 mots)\n\nRéponds uniquement en JSON : {\"title\": \"...\", \"alt\": \"...\", \"caption\": \"...\"}",
"ai.imageAnalysis.user": "Analyse cette image. Réponds en JSON en français."
}

View File

@@ -78,5 +78,7 @@
"render.month.9": "settembre",
"render.month.10": "ottobre",
"render.month.11": "novembre",
"render.month.12": "dicembre"
"render.month.12": "dicembre",
"ai.imageAnalysis.system": "Generi metadati per immagini. Scrivi tutti i valori in italiano.\n\nRegole:\n- \"title\": titolo descrittivo breve (3-8 parole)\n- \"alt\": descrizione fattuale di ciò che è visibile (5-12 parole). Nessuna interpretazione. Nessun prefisso \"Immagine di\".\n- \"caption\": didascalia blog coinvolgente (5-20 parole)\n\nRispondi solo con JSON: {\"title\": \"...\", \"alt\": \"...\", \"caption\": \"...\"}",
"ai.imageAnalysis.user": "Analizza questa immagine. Rispondi con JSON in italiano."
}