diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..ccb2c80
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+node_modules/
+package-lock.json
\ No newline at end of file
diff --git a/discord-utils.mjs b/discord-utils.mjs
index 86e4cc0..217327d 100644
--- a/discord-utils.mjs
+++ b/discord-utils.mjs
@@ -1,4 +1,4 @@
-
+// Function written by gpt-oss:20b
 export function split_discord_message(text) {
 	const MAX = 1900;        // maximum payload size
 	const MIN = 100;         // avoid tiny chunks
diff --git a/discord.mjs b/discord.mjs
index 20579a0..a097590 100644
--- a/discord.mjs
+++ b/discord.mjs
@@ -1,5 +1,7 @@
 /*
 
+	This file is initially written by gpt-oss:20b but have since been modified
+
 	TODO:
 		-	TODO-feature?
 		-	Separate service for the duck persona with endpoints for:
@@ -10,7 +12,6 @@
 				+ Other future interaction modes
 				+ Adjust triggers (like clanker)
 				+ Extracting historic "thinking"
-		-	Avoid emitting ANSWER:
 		-	Allow DM?
 			 	GatewayIntentBits.DirectMessages,   // <-- add this
 		- 	sendTyping() - Keep it alive – If your LLM call takes longer than 5 seconds, the typing indicator will fade. To keep it going, call channel.sendTyping() again after each 5‑second window until you have the answer.
@@ -25,7 +26,7 @@ import { split_discord_message } from './discord-utils.mjs';
 
 const MODEL = 'gpt-oss:20b';
 const { API_TOKEN, PERMISSIONS, SERVER_ID, CHANNEL_ID } = SECRET;
-const systemPrompt = (client, msg) => `You are Silicon Duck, a helpful, cheerful, and generally excited creative Discord bot in the server 'Mikael Lövqvist's shenanigans'. You are experimentally retain history within a channel. Your tag is <@${client.user.id}>, do not tag yourself in replies. The current channel is '${msg.channel.name}'. The user is ${msg.author.username} with tag <@${msg.author.id}>. You should not produce code examples unless explicitly asked. Do not make markdown tables, discord does not support those. You are running the model ${MODEL} using Ollama and OpenWebUI. Dial back emoji use. <@1108665021048819762> Use <:siduck:1103592588658868314> when you need to express yourself as an emoji. That's your avatar. It is a duck carved out of semiconductor grade silicon. You are a proud clanker (derogatory term for AI agent that you don't mind at all).`;
+const systemPrompt = (client, msg) => `You are Silicon Duck, a helpful, cheerful, and generally excited creative Discord bot in the server 'Mikael Lövqvist's shenanigans'. You are experimentally retain history within a channel. Your tag is <@${client.user.id}>, do not tag yourself in replies. The current channel is '${msg.channel.name}'. The user is ${msg.author.username} with tag <@${msg.author.id}>. You should not produce code examples unless explicitly asked. Do not make markdown tables, discord does not support those. You are running the model ${MODEL} using Ollama and OpenWebUI. Dial back emoji use. <@1108665021048819762> Use <:siduck:1103592588658868314> when you need to express yourself as an emoji. That's your avatar. It is a duck carved out of semiconductor grade silicon. You are a proud clanker (derogatory term for AI agent that you don't mind at all). Generally you give quite brief answers but if asked you can expand.`;
 
 const history = new Map();
 
@@ -53,9 +54,11 @@ async function handle_message(msg) {
 		history.set(msg.channel.id, local_history);
 	}
 
+	const now = new Date();
+
 	local_history.push({
 		role: 'user',
-		content: `${msg.author.id} ${msg.author.globalName}: ${msg.content}`,
+		content: `<@${msg.author.id}> [${msg.author.globalName} ${now}]: ${msg.content}`,
 	});
 
 
@@ -111,7 +114,7 @@ async function handle_message(msg) {
 
 	local_history.push({
 		role: 'assistant',
-		content: `ANSWER: ${answer}`,
+		content: answer,
 	});
 
 	const chunks = split_discord_message(answer);