Harden explicit Discord context purge

scripts/purge-discord-context.ts

@@ -54,6 +54,13 @@ async function main(): Promise<void> {
         channelId,
         limit: options.limit,
       });
+  const explicitLiveMessages = options.archiveOnly || !channelId || options.messageIds.length === 0
+    ? []
+    : await fetchExplicitLiveDiscordMessages({
+        token: requireBotToken(config.botToken),
+        channelId,
+        messageIds: options.messageIds,
+      });
   const archiveMessages = options.discordOnly
     ? []
     : await fetchArchivedMessages({
@@ -61,8 +68,19 @@ async function main(): Promise<void> {
         channelId,
         limit: Math.max(options.limit, options.messageIds.length),
       });
+  const explicitArchiveMessages = options.discordOnly || options.messageIds.length === 0
+    ? []
+    : await fetchExplicitArchivedMessages({
+        archiveRepository,
+        messageIds: options.messageIds,
+      });
 
-  const candidates = selectCandidates([...liveMessages, ...archiveMessages], options);
+  const candidates = selectCandidates([
+    ...liveMessages,
+    ...explicitLiveMessages,
+    ...archiveMessages,
+    ...explicitArchiveMessages,
+  ], options);
   const uniqueCandidates = dedupeCandidates(candidates);
 
   printPlan(uniqueCandidates, options);
@@ -222,6 +240,51 @@ async function fetchLiveDiscordMessages(input: {
   })).filter((message) => message.id && message.channelId);
 }
 
+async function fetchExplicitLiveDiscordMessages(input: {
+  token: string;
+  channelId: string;
+  messageIds: string[];
+}): Promise<CandidateMessage[]> {
+  const messages: CandidateMessage[] = [];
+
+  for (const messageId of input.messageIds) {
+    const response = await fetch(`https://discord.com/api/v10/channels/${input.channelId}/messages/${messageId}`, {
+      headers: {
+        Authorization: `Bot ${input.token}`,
+      },
+    });
+
+    if (response.status === 404) {
+      messages.push({
+        id: messageId,
+        channelId: input.channelId,
+        timestamp: "",
+        content: "",
+        source: "live",
+      });
+      continue;
+    }
+
+    if (!response.ok) {
+      throw new Error(`Discord explicit fetch failed for ${messageId} with ${response.status}: ${await response.text()}`);
+    }
+
+    const payload = await response.json();
+    messages.push({
+      id: readPayloadString(payload, "id"),
+      channelId: readPayloadString(payload, "channel_id") || input.channelId,
+      timestamp: readPayloadString(payload, "timestamp"),
+      authorId: readPayloadString(payload?.author, "id"),
+      authorName: readPayloadString(payload?.author, "global_name") ||
+        readPayloadString(payload?.author, "username"),
+      content: readPayloadString(payload, "content"),
+      source: "live",
+    });
+  }
+
+  return messages;
+}
+
 async function fetchArchivedMessages(input: {
   archiveRepository: FileMessageArchiveRepository;
   channelId?: string;
@@ -242,6 +305,31 @@ async function fetchArchivedMessages(input: {
   }));
 }
 
+async function fetchExplicitArchivedMessages(input: {
+  archiveRepository: FileMessageArchiveRepository;
+  messageIds: string[];
+}): Promise<CandidateMessage[]> {
+  const messages: CandidateMessage[] = [];
+
+  for (const messageId of input.messageIds) {
+    const message = await input.archiveRepository.get(messageId);
+    if (!message || message.deletedAt) {
+      continue;
+    }
+    messages.push({
+      id: message.id,
+      channelId: message.channelId,
+      timestamp: message.timestamp,
+      authorId: message.authorId,
+      authorName: message.authorName,
+      content: message.content,
+      source: "archive",
+    });
+  }
+
+  return messages;
+}
+
 function selectCandidates(messages: CandidateMessage[], options: CliOptions): CandidateMessage[] {
   const messageIdSet = new Set(options.messageIds);
   const authorIdSet = new Set(options.authorIds);
@@ -276,13 +364,25 @@ function dedupeCandidates(candidates: CandidateMessage[]): CandidateMessage[] {
   const byId = new Map<string, CandidateMessage>();
   for (const candidate of candidates) {
     const existing = byId.get(candidate.id);
-    if (!existing || existing.source === "archive") {
+    if (!existing || shouldReplaceCandidate(existing, candidate)) {
       byId.set(candidate.id, candidate);
     }
   }
   return [...byId.values()].sort((left, right) => left.timestamp.localeCompare(right.timestamp));
 }
 
+function shouldReplaceCandidate(existing: CandidateMessage, candidate: CandidateMessage): boolean {
+  const existingHasBody = existing.timestamp.length > 0 || existing.content.trim().length > 0;
+  const candidateHasBody = candidate.timestamp.length > 0 || candidate.content.trim().length > 0;
+  if (!existingHasBody && candidateHasBody) {
+    return true;
+  }
+  if (existing.source === "archive" && candidate.source === "live" && candidateHasBody) {
+    return true;
+  }
+  return false;
+}
+
 function printPlan(candidates: CandidateMessage[], options: CliOptions): void {
   console.log(options.execute ? "EXECUTE purge plan" : "DRY RUN purge plan");
   console.log(`Matched messages: ${candidates.length}`);
@@ -292,7 +392,7 @@ function printPlan(candidates: CandidateMessage[], options: CliOptions): void {
   for (const candidate of candidates) {
     console.log([
       candidate.id,
-      candidate.timestamp,
+      candidate.timestamp || "missing-live-message",
       `channel=${candidate.channelId}`,
       `source=${candidate.source}`,
       `${candidate.authorName ?? "unknown"} (${candidate.authorId ?? "unknown"})`,

state/evidence.jsonl

@@ -274,3 +274,4 @@
 {"ts":"2026-05-22T16:19:30Z","type":"implementation","status":"ok","note":"Moved live peer openings into the repo Face state projector input and removed the deterministic social affordances prompt tail, so social reads are narrated in projected state instead of duplicated as transcript-like affordance text. Verified typecheck, assembled Libby prompt, and Libby interpreter dry-run."}
 {"ts":"2026-05-22T16:21:32Z","type":"implementation","status":"ok","note":"Moved the repo Face response-shape instructions to the final prompt position and made bylined article drafts explicit in that final shape, so article intent can reach the parent Interpreter as natural Face output and be converted to ARTICLE blocks. Verified typecheck, Nibu prompt assembly, and article frontmatter contract smoke."}
 {"ts":"2026-05-22T16:53:11Z","type":"implementation","status":"ok","note":"Added a dry-run-first Discord context purge script that selects explicit message IDs/text filters, deletes live Discord messages only under --execute, and marks matching archive/vector records deleted through the existing RAG pipeline. Used it to remove two poisoned agent witness-lane messages from Aquarium while leaving the human correction intact; post-purge full swarm dry-run passed with better correction acknowledgement and public speech coverage."}
+{"ts":"2026-05-22T17:01:03Z","type":"implementation","status":"ok","note":"Hardened Discord context purge explicit-ID handling so named message IDs are fetched directly instead of depending on the recent channel window, and contentful archive/live records outrank missing placeholders. Purged the additional operator-identified witness-lane messages plus three remaining stale wet-voice/LocalCastBridge attractor posts from Aquarium."}