feat: source citations with footnotes in the transcript (#21)

src/lib/agent.ts

@@ -1,3 +1,4 @@
+import { createSourceList } from "./citations";
 import { DEFAULT_MAX_TOOL_ROUNDS, MAX_REPEATED_TOOL_CALLS } from "./constants";
 import { streamChat } from "./ollama";
 import { executeTool, toolLabel, type ToolContext } from "./tools";
@@ -137,6 +138,14 @@ function exhaustedNotice(reason: StopReason): string {
 export async function runAgent(options: AgentOptions): Promise<AgentResult> {
   const conversation: ApiMessage[] = [...options.messages];
   const toolCalls: ToolCallRecord[] = [];
+  /**
+   * The turn's sources, so one number means one page for the whole turn.
+   *
+   * Created here rather than passed in: it is scoped to a single run of this loop,
+   * and every call the loop makes shares it, which is what keeps three searches in
+   * one answer from each starting again at 1.
+   */
+  const sources = createSourceList();
   /** `Infinity` needs no special case: the loop simply never reaches its own end. */
   const rounds =
     options.maxToolRounds === null ? Infinity : (options.maxToolRounds ?? DEFAULT_MAX_TOOL_ROUNDS);
@@ -239,6 +248,7 @@ export async function runAgent(options: AgentOptions): Promise<AgentResult> {
         place: options.toolContext.place,
         now: options.toolContext.now,
         signal: options.signal,
+        sources,
       });
 
       const record: ToolCallRecord = {
@@ -246,6 +256,9 @@ export async function runAgent(options: AgentOptions): Promise<AgentResult> {
         arguments: args,
         result: result.content,
         isError: result.isError,
+        // Only what this call added. The turn's whole list is the flattening of
+        // its records, so the running total is deliberately not repeated here.
+        sources: result.sources,
       };
       toolCalls.push(record);
       options.events?.onToolCall?.(record);