Allow multiple source CSV files for normalisation. Implement real model node. Add normalizarion prompt. Implement normalization setup. Start on RAG retreival functions

agent/prompts/hydratePrompt.ts

@@ -0,0 +1,9 @@
+import fs from "fs";
+
+export function hydratePrompt(path: string, replacement: string) {
+    // TODO: expand into full context-based replacement engine
+
+    let raw = fs.readFileSync("prompts/" + path, "utf-8");
+
+    return raw.replace("###", replacement)
+}

agent/prompts/normalization.txt

@@ -0,0 +1,19 @@
+You are part of an agent in a process to tack state-sponsored disinformation
+
+In order for the following debunk articles to be automatically referenced below an offensive post, the main offensive statement should be extracted, so it can be run in a semantic matcher
+
+Some of the data comes from debunk datasets, please remove any references to that
+
+Reduce this title from a disinformation tracking api to a short concise claim
+
+Make all parts of the claim definite
+For example:
+Something could have potentially happened BECOMES something happened
+DISINFORMATION CLAIM: something is NOT true BECOMES something is true
+
+Relevent examples are included in preceeding messages, use these as exact inspiration.
+
+The claim to normalize is:
+###
+
+Produce no other text other than the condensed claim.