AI-powered roleplay with characters that remember, adapt, and feel real. Deep personality systems, persistent memory, and narrative intelligence. All stored locally. Your data never leaves your machine.
Built by a roleplayer, for roleplayers. Every feature exists because something was missing from existing tools.
Bring your own API key. Connect to OpenAI, Anthropic, OpenRouter, NanoGPT, Chutes, or any OpenAI-compatible proxy. Switch models mid-session with a fallback if your primary fails.
Type ((OOC: keep it short)) anywhere in your message. A free local AI model (Qwen 3B, runs offline on your GPU) extracts it as a private instruction and ships clean in-character text to your cloud proxy. Your tokens go to roleplay, not housekeeping.
Go beyond text dumps. Identity, appearance, personality sliders, background, connections. Characters with real depth.
Trigger-based context injection. Your world's lore loads only when relevant, keeping context focused and responses sharp.
Runs in the background while you play. No "save" button, no manual checkpoints, no forgetting to summarize before bed. LoreWright watches the chat and summarizes automatically the moment it crosses the token threshold you set — then keeps the summary fresh in every message. You just write.
LoreWright detects whether your input is passive, interactive, or decision-driven, then adjusts pacing automatically. No manual mode switching.
Split your system prompt into named, toggleable pieces. Flip a directive on per session with one click — No Clichés, Slow Burn, Plot Driver, Jailbreak, and 30 more shipped built-in, or write your own. Same flexibility as a giant custom prompt without the wall-of-text mess.
Define who you are in the story. Avatar, name, pronouns, description. Switch personas between sessions.
Don't like a response? Reroll for alternatives. Navigate between variants, lock your favorite, delete the rest.
Most platforms shove everything into one text box and hope the AI figures it out. LoreWright gives characters real structure: identity, appearance, personality spectrums, sexuality, background, and connections. The AI doesn't guess who your character is. It knows.
Long conversations lose context. LoreWright doesn't. It automatically summarizes what's happened and keeps it in every message, so your character stays consistent from the first line to the last. No more repeating yourself because the AI "forgot."
Build lorebooks with trigger-based entries that inject context when mentioned, or pin entries to be always active for core world knowledge. Keep your world consistent without burning your entire token budget. Group lorebooks into reusable world packages.
You don't need to copy a 2000-word jailbreak off Reddit to make the AI stop speaking in your voice. LoreWright ships with battle-tested rules baked into every prompt — the same fixes that everyone keeps re-inventing, already there from message one:
You can still override the system prompt per session if you want full control. You usually won't need to.
A note on AI behavior: LoreWright positions every rule, directive, and instruction where models pay the most attention — but prompts are not contracts. The AI has the final word, and larger / smarter models follow guardrails more reliably than smaller ones. We tune what we can control; you pick the proxy.
Not just another chatbot wrapper.
Toggleable behaviors across style, story, system, and roleplay categories
Passive, interactive, and decision-driven pacing, detected and adjusted automatically
We never see your conversations. Your prompts go directly to your chosen AI provider. Not through us. Ever.
Add your OpenAI, Anthropic, or any compatible API key. Your keys stay encrypted on your machine.
Create one from scratch, or import any Character Card V2 file (works with Chub, JanitorAI, Risu, SillyTavern exports). Add personality, background, and let them come alive.
That's it. Memory, lorebooks, narrative pacing, all work automatically. Just write.
LoreWright is in closed beta. Limited spots. Help shape the product and get a discount on launch.