v1.11.6: doom-loop guard (3 identical tool calls aborts recursion)

apps/server/src/services/__tests__/doom-loop.test.ts

@@ -0,0 +1,130 @@
+import { describe, it, expect } from 'vitest';
+import { DOOM_LOOP_THRESHOLD, detectDoomLoop } from '../inference.js';
+import type { ToolCall } from '../../types/api.js';
+
+// ---- fixture ----------------------------------------------------------------
+// Tiny helper. `id` is required on ToolCall but irrelevant to detection —
+// detectDoomLoop compares name + JSON.stringify(args). Counter-based id keeps
+// each call unique so we don't accidentally test id-based equality.
+
+let counter = 0;
+function mkCall(name: string, args: Record<string, unknown> = {}): ToolCall {
+  counter += 1;
+  return { id: `c${counter}`, name, args };
+}
+
+// ---- below-threshold -------------------------------------------------------
+
+describe('detectDoomLoop — below threshold', () => {
+  it('returns null for an empty array', () => {
+    expect(detectDoomLoop([])).toBeNull();
+  });
+
+  it('returns null when fewer than DOOM_LOOP_THRESHOLD calls exist', () => {
+    // 2 < 3 — sliding-window can't form even if both match.
+    const a = mkCall('view_file', { path: 'a.ts' });
+    const b = mkCall('view_file', { path: 'a.ts' });
+    expect(detectDoomLoop([a, b])).toBeNull();
+  });
+});
+
+// ---- positive detection ----------------------------------------------------
+
+describe('detectDoomLoop — positive matches', () => {
+  it('returns name + args when exactly DOOM_LOOP_THRESHOLD identical calls land', () => {
+    const calls = [
+      mkCall('grep', { pattern: 'TODO', path: 'src' }),
+      mkCall('grep', { pattern: 'TODO', path: 'src' }),
+      mkCall('grep', { pattern: 'TODO', path: 'src' }),
+    ];
+    const result = detectDoomLoop(calls);
+    expect(result).not.toBeNull();
+    expect(result!.name).toBe('grep');
+    expect(result!.args).toEqual({ pattern: 'TODO', path: 'src' });
+  });
+
+  it('matches sliding window — last DOOM_LOOP_THRESHOLD match even with earlier non-matching calls', () => {
+    // 4 calls: first differs, last 3 are identical → fire.
+    const calls = [
+      mkCall('list_dir', { path: '/' }),
+      mkCall('view_file', { path: 'a.ts' }),
+      mkCall('view_file', { path: 'a.ts' }),
+      mkCall('view_file', { path: 'a.ts' }),
+    ];
+    const result = detectDoomLoop(calls);
+    expect(result).not.toBeNull();
+    expect(result!.name).toBe('view_file');
+  });
+
+  it('matches identical empty-args calls (defense against {} !== {} reference bug)', () => {
+    // JSON.stringify on two distinct {} both produce '{}'. Confirms the
+    // detector uses value-equality not reference-equality.
+    const calls = [mkCall('ping', {}), mkCall('ping', {}), mkCall('ping', {})];
+    expect(detectDoomLoop(calls)).not.toBeNull();
+  });
+
+  it('matches calls with nested args of equal shape', () => {
+    // Deep-equal via JSON.stringify. If the model emits the same nested
+    // object three times, that's still a loop.
+    const nested = { filter: { glob: '*.ts', case: 'sensitive' }, limit: 50 };
+    const calls = [
+      mkCall('find_files', { ...nested }),
+      mkCall('find_files', { ...nested }),
+      mkCall('find_files', { ...nested }),
+    ];
+    expect(detectDoomLoop(calls)).not.toBeNull();
+  });
+});
+
+// ---- negative detection ----------------------------------------------------
+
+describe('detectDoomLoop — negative cases', () => {
+  it('returns null when 3 calls share name but differ in args', () => {
+    const calls = [
+      mkCall('view_file', { path: 'a.ts' }),
+      mkCall('view_file', { path: 'b.ts' }),
+      mkCall('view_file', { path: 'c.ts' }),
+    ];
+    expect(detectDoomLoop(calls)).toBeNull();
+  });
+
+  it('returns null when 3 calls share args but differ in name', () => {
+    const calls = [
+      mkCall('view_file', { path: 'a.ts' }),
+      mkCall('grep', { path: 'a.ts' }),
+      mkCall('list_dir', { path: 'a.ts' }),
+    ];
+    expect(detectDoomLoop(calls)).toBeNull();
+  });
+
+  it('returns null when the FIRST three of four match but the latest differs', () => {
+    // Critical sliding-window edge: detector must ONLY look at the last
+    // DOOM_LOOP_THRESHOLD entries. Earlier matches don't count if the
+    // model has since moved on.
+    const calls = [
+      mkCall('grep', { pattern: 'X' }),
+      mkCall('grep', { pattern: 'X' }),
+      mkCall('grep', { pattern: 'X' }),
+      mkCall('view_file', { path: 'a.ts' }),
+    ];
+    expect(detectDoomLoop(calls)).toBeNull();
+  });
+
+  it('returns null when args have same keys but different values', () => {
+    const calls = [
+      mkCall('grep', { pattern: 'TODO', path: 'src' }),
+      mkCall('grep', { pattern: 'TODO', path: 'src' }),
+      mkCall('grep', { pattern: 'TODO', path: 'apps' }),
+    ];
+    expect(detectDoomLoop(calls)).toBeNull();
+  });
+});
+
+// ---- threshold contract ----------------------------------------------------
+
+describe('DOOM_LOOP_THRESHOLD', () => {
+  it('is a positive integer (the public contract — tests assume 3)', () => {
+    expect(DOOM_LOOP_THRESHOLD).toBeGreaterThan(0);
+    expect(Number.isInteger(DOOM_LOOP_THRESHOLD)).toBe(true);
+  });
+});

apps/server/src/services/inference.ts

@@ -54,6 +54,36 @@ function resolveToolBudget(agent: Agent | null): number {
 const CAP_HIT_SUMMARY_NOTE = (limit: number) =>
   `You've reached the tool budget (${limit} calls). Produce the best answer you can with what you have. Do not call more tools.`;
 
+// v1.11.6: doom-loop guard. When the model calls the same tool with the
+// same arguments DOOM_LOOP_THRESHOLD times in a row within one user-message
+// turn, abort the recursion and run the same wrap-up summary path as the
+// cap-hit case. Ported from opencode (DOOM_LOOP_THRESHOLD in
+// session/processor.ts). Threshold of 3 is the smallest value that doesn't
+// false-positive on a model that retries once after a transient error.
+export const DOOM_LOOP_THRESHOLD = 3;
+
+const DOOM_LOOP_NOTE = (name: string) =>
+  `You called ${name} with the same arguments ${DOOM_LOOP_THRESHOLD} times in a row. Stop calling it. Produce the best answer you can with what you have.`;
+
+// Returns the name + args of the looping tool when the LAST
+// DOOM_LOOP_THRESHOLD entries in `recentToolCalls` are identical (same name
+// AND deep-equal args via JSON.stringify). Returns null otherwise.
+// Pure; exported for unit-test access.
+export function detectDoomLoop(
+  recentToolCalls: ToolCall[],
+): { name: string; args: Record<string, unknown> } | null {
+  if (recentToolCalls.length < DOOM_LOOP_THRESHOLD) return null;
+  const last = recentToolCalls.slice(-DOOM_LOOP_THRESHOLD);
+  const ref = last[0]!;
+  const refArgs = JSON.stringify(ref.args);
+  for (let i = 1; i < last.length; i++) {
+    const tc = last[i]!;
+    if (tc.name !== ref.name) return null;
+    if (JSON.stringify(tc.args) !== refArgs) return null;
+  }
+  return { name: ref.name, args: ref.args };
+}
+
 function isCapHitSentinel(m: Message): boolean {
   return (
     m.role === 'system' &&
@@ -63,6 +93,22 @@ function isCapHitSentinel(m: Message): boolean {
   );
 }
 
+// v1.11.6: parallel predicate. Same UI-only semantics as cap-hit sentinels —
+// never sent to the LLM (filtered by buildMessagesPayload through the
+// isAnySentinel check below).
+function isDoomLoopSentinel(m: Message): boolean {
+  return (
+    m.role === 'system' &&
+    m.metadata !== null &&
+    typeof m.metadata === 'object' &&
+    (m.metadata as { kind?: unknown }).kind === 'doom_loop'
+  );
+}
+
+function isAnySentinel(m: Message): boolean {
+  return isCapHitSentinel(m) || isDoomLoopSentinel(m);
+}
+
 export interface InferenceFrame {
   type:
     | 'message_started'
@@ -203,11 +249,11 @@ export function buildMessagesPayload(
       out.push({ role: 'system', content: m.content });
       continue;
     }
-    // v1.8.2: cap-hit sentinels are UI-only — never send them to the LLM. The
-    // synthetic "you've reached the tool budget" note lives only inside the
-    // summary call's messages array and is never persisted, so on Continue
-    // the model resumes with a clean context.
-    if (isCapHitSentinel(m)) continue;
+    // v1.8.2 / v1.11.6: cap-hit and doom-loop sentinels are UI-only — never
+    // send them to the LLM. The synthetic instruction note lives only inside
+    // the summary call's messages array and is never persisted, so on a
+    // follow-up turn the model resumes with a clean context.
+    if (isAnySentinel(m)) continue;
     if (m.role === 'assistant' && m.status === 'streaming') continue;
     if (m.role === 'assistant' && m.status === 'cancelled') continue;
     if (m.role === 'tool') {
@@ -608,6 +654,11 @@ interface TurnArgs {
   // resolved budget at the top of each turn. Replaces the older `depth`
   // counter (which counted iterations, not invocations).
   toolsUsed: number;
+  // v1.11.6: ordered tool calls executed in this user-message turn (across
+  // recursive runAssistantTurn invocations). Reset to [] at user-message
+  // boundaries by runInference, same as toolsUsed. Doom-loop check at the
+  // top of runAssistantTurn slices the last DOOM_LOOP_THRESHOLD entries.
+  recentToolCalls: ToolCall[];
   signal: AbortSignal | undefined;
 }
 
@@ -910,6 +961,11 @@ async function executeToolPhase(
     // One assistant message can emit multiple tool_calls, so we add the run
     // count, not 1. The next turn's budget check sees the cumulative total.
     toolsUsed: toolsUsed + result.toolCalls.length,
+    // v1.11.6: append the just-executed tool calls to the per-turn history
+    // so the next runAssistantTurn's doom-loop check can see them. We don't
+    // cap the array length here — per-turn budgets keep it bounded
+    // (typically <30 entries), and slicing happens inside detectDoomLoop.
+    recentToolCalls: [...args.recentToolCalls, ...result.toolCalls],
     signal,
   });
 }
@@ -1029,6 +1085,17 @@ async function runAssistantTurn(
     return;
   }
 
+  // v1.11.6: doom-loop guard. Detected BEFORE the budget cap (the model can
+  // burn through 3 identical calls long before the 15-call budget fires).
+  // Same in-flight-slot-reuse pattern as runCapHitSummary — wrap-up reply
+  // lands in args.assistantMessageId, then a doom_loop sentinel is inserted
+  // to make the abort visible in the chat history.
+  const loop = detectDoomLoop(args.recentToolCalls);
+  if (loop) {
+    await runDoomLoopSummary(ctx, args, session, project, history, agent, loop);
+    return;
+  }
+
   const messages = buildMessagesPayload(session, project, history, agent);
 
   const state: StreamPhaseState = { accumulated: '', startedAt: null };
@@ -1059,7 +1126,16 @@ export async function runInference(
   // continue) starts with a clean budget. Tool-call accumulation across
   // Continue invocations is what the hard ceiling guards against, not the
   // per-call budget.
-  return runAssistantTurn(ctx, { sessionId, chatId, assistantMessageId, toolsUsed: 0, signal });
+  // v1.11.6: recentToolCalls also resets — doom-loop detection is scoped
+  // to a single user-message turn, so a Continue starts with no history.
+  return runAssistantTurn(ctx, {
+    sessionId,
+    chatId,
+    assistantMessageId,
+    toolsUsed: 0,
+    recentToolCalls: [],
+    signal,
+  });
 }
 
 // v1.8.2: cap-hit summary flow. Called instead of erroring when the loop
@@ -1318,6 +1394,250 @@ async function insertCapHitSentinel(
   });
 }
 
+// v1.11.6: doom-loop wrap-up. Mirrors runCapHitSummary structurally — same
+// in-flight-slot reuse, same tools-disabled streaming-summary call, same
+// post-finalize sentinel insert + chat_status drop. Differences:
+//   - synthetic note text comes from DOOM_LOOP_NOTE (names the looping tool)
+//   - sentinel metadata is { kind: 'doom_loop', tool_name, args, threshold }
+//     and has no Continue affordance (manual retry would just re-loop)
+//   - chat_status error path uses reason: 'doom_loop_summary_failed'
+// Kept as a clone rather than refactored into a shared helper because the
+// two summary paths still differ in error reason + sentinel shape; a third
+// sentinel would justify factoring out runWrapUpSummary(opts).
+async function runDoomLoopSummary(
+  ctx: InferenceContext,
+  args: TurnArgs,
+  session: Session,
+  project: Project,
+  history: Message[],
+  agent: Agent | null,
+  loop: { name: string; args: Record<string, unknown> },
+): Promise<void> {
+  const { sessionId, chatId, assistantMessageId, signal } = args;
+
+  const messages = buildMessagesPayload(session, project, history, agent);
+  messages.push({ role: 'system', content: DOOM_LOOP_NOTE(loop.name) });
+
+  const startedRow = await ctx.sql<{ started_at: string }[]>`
+    UPDATE messages
+    SET started_at = clock_timestamp()
+    WHERE id = ${assistantMessageId}
+    RETURNING started_at
+  `;
+  const startedAt = startedRow[0]?.started_at ?? null;
+
+  ctx.publish(sessionId, {
+    type: 'message_started',
+    message_id: assistantMessageId,
+    chat_id: chatId,
+    role: 'assistant',
+  });
+
+  let accumulated = '';
+  let pendingFlushTimer: NodeJS.Timeout | null = null;
+  let flushPromise: Promise<unknown> = Promise.resolve();
+  const flushNow = () => {
+    if (pendingFlushTimer) {
+      clearTimeout(pendingFlushTimer);
+      pendingFlushTimer = null;
+    }
+    const snapshot = accumulated;
+    flushPromise = flushPromise.then(() =>
+      ctx.sql`UPDATE messages SET content = ${snapshot} WHERE id = ${assistantMessageId}`
+    );
+  };
+  const scheduleFlush = () => {
+    if (pendingFlushTimer) return;
+    pendingFlushTimer = setTimeout(() => {
+      pendingFlushTimer = null;
+      flushNow();
+    }, DB_FLUSH_INTERVAL_MS);
+  };
+
+  let summaryOk = false;
+  let summarySoftCancelled = false;
+  let summaryError: string | null = null;
+  let result: StreamResult | null = null;
+  try {
+    result = await streamCompletion(
+      ctx,
+      session.model,
+      messages,
+      { tools: null, temperature: agent?.temperature },
+      (delta) => {
+        accumulated += delta;
+        ctx.publish(sessionId, {
+          type: 'delta',
+          message_id: assistantMessageId,
+          chat_id: chatId,
+          content: delta,
+        });
+        scheduleFlush();
+      },
+      signal,
+    );
+    summaryOk = true;
+  } catch (err) {
+    if (err instanceof Error && err.name === 'AbortError') {
+      summarySoftCancelled = true;
+    } else {
+      summaryError = err instanceof Error ? err.message : String(err);
+    }
+  } finally {
+    if (pendingFlushTimer) {
+      clearTimeout(pendingFlushTimer);
+      pendingFlushTimer = null;
+    }
+    await flushPromise;
+  }
+
+  if (summaryOk && result) {
+    const mctx = await modelContext.getModelContext(session.model);
+    const nCtx = mctx?.n_ctx ?? null;
+    const [updated] = await ctx.sql<
+      { tokens_used: number | null; ctx_used: number | null; ctx_max: number | null; finished_at: string | null }[]
+    >`
+      UPDATE messages
+      SET content = ${result.content},
+          status = 'complete',
+          tokens_used = ${result.completionTokens},
+          ctx_used = ${result.promptTokens},
+          ctx_max = ${nCtx},
+          finished_at = clock_timestamp()
+      WHERE id = ${assistantMessageId}
+      RETURNING tokens_used, ctx_used, ctx_max, finished_at
+    `;
+    ctx.publish(sessionId, {
+      type: 'message_complete',
+      message_id: assistantMessageId,
+      chat_id: chatId,
+      tokens_used: updated?.tokens_used ?? null,
+      ctx_used: updated?.ctx_used ?? null,
+      ctx_max: updated?.ctx_max ?? null,
+      started_at: startedAt,
+      finished_at: updated?.finished_at ?? null,
+      model: session.model,
+    });
+  } else if (summarySoftCancelled) {
+    await ctx.sql`
+      UPDATE messages
+      SET content = ${accumulated},
+          status = 'cancelled',
+          finished_at = clock_timestamp()
+      WHERE id = ${assistantMessageId}
+    `;
+    ctx.publish(sessionId, {
+      type: 'message_complete',
+      message_id: assistantMessageId,
+      chat_id: chatId,
+    });
+  } else {
+    // Doom-loop summary failure reuses the existing summary_after_cap_failed
+    // error reason — the ErrorReason union is shared between sentinel paths
+    // and the UI surfaces a generic "summary failed" line for both. We don't
+    // add a new reason code because the user-visible failure mode is the
+    // same (model gave up mid-summary). Sentinel below still fires.
+    const errMeta: MessageMetadata = {
+      kind: 'error',
+      error_reason: 'summary_after_cap_failed',
+      error_text: summaryError ?? 'doom-loop summary failed',
+    };
+    await ctx.sql`
+      UPDATE messages
+      SET content = ${accumulated},
+          status = 'failed',
+          finished_at = clock_timestamp(),
+          metadata = ${ctx.sql.json(errMeta as never)}
+      WHERE id = ${assistantMessageId}
+    `;
+    ctx.publish(sessionId, {
+      type: 'error',
+      message_id: assistantMessageId,
+      chat_id: chatId,
+      error: summaryError ?? 'doom-loop summary failed',
+      reason: 'summary_after_cap_failed',
+    });
+  }
+
+  const [sessRow] = await ctx.sql<{ project_id: string; name: string; updated_at: string }[]>`
+    UPDATE sessions SET updated_at = clock_timestamp()
+    WHERE id = ${sessionId}
+    RETURNING project_id, name, updated_at
+  `;
+  ctx.publishUser({
+    type: 'session_updated',
+    session_id: sessionId,
+    project_id: sessRow!.project_id,
+    name: sessRow!.name,
+    updated_at: sessRow!.updated_at,
+  });
+
+  await insertDoomLoopSentinel(ctx, sessionId, chatId, loop);
+
+  if (summaryOk || summarySoftCancelled) {
+    ctx.publishUser({ type: 'chat_status', chat_id: chatId, status: 'idle', at: new Date().toISOString() });
+  } else {
+    ctx.publishUser({
+      type: 'chat_status',
+      chat_id: chatId,
+      status: 'error',
+      at: new Date().toISOString(),
+      reason: 'summary_after_cap_failed',
+    });
+  }
+
+  ctx.log.info(
+    { sessionId, chatId, assistantMessageId, loopedTool: loop.name, summaryOk, summaryCancelled: summarySoftCancelled },
+    'inference doom-loop summary finished',
+  );
+}
+
+async function insertDoomLoopSentinel(
+  ctx: InferenceContext,
+  sessionId: string,
+  chatId: string,
+  loop: { name: string; args: Record<string, unknown> },
+): Promise<void> {
+  // No hard-ceiling / can-continue logic here — doom-loop is a different
+  // failure mode from cap-hit. Continuing would re-trigger the loop with
+  // the same tools available; the user needs to restate their question
+  // or switch agents instead.
+  const metadata: MessageMetadata = {
+    kind: 'doom_loop',
+    tool_name: loop.name,
+    args: loop.args,
+    threshold: DOOM_LOOP_THRESHOLD,
+  };
+  const content = `Detected ${DOOM_LOOP_THRESHOLD} identical calls to ${loop.name}. Stopping the tool-call loop. Produce the best answer you can with what you have.`;
+
+  const [row] = await ctx.sql<{ id: string }[]>`
+    INSERT INTO messages (session_id, chat_id, role, content, status, created_at, metadata)
+    VALUES (${sessionId}, ${chatId}, 'system', ${content}, 'complete', clock_timestamp(), ${ctx.sql.json(metadata as never)})
+    RETURNING id
+  `;
+
+  // Standard frame sequence — same as cap-hit sentinel — so
+  // useSessionStream's reducer appends the row via the existing path.
+  ctx.publish(sessionId, {
+    type: 'message_started',
+    message_id: row!.id,
+    chat_id: chatId,
+    role: 'system',
+  });
+  ctx.publish(sessionId, {
+    type: 'delta',
+    message_id: row!.id,
+    chat_id: chatId,
+    content,
+  });
+  ctx.publish(sessionId, {
+    type: 'message_complete',
+    message_id: row!.id,
+    chat_id: chatId,
+    metadata,
+  });
+}
+
 interface InferenceRegistration {
   controller: AbortController;
   completed: Promise<void>;

apps/server/src/types/api.ts

@@ -128,9 +128,11 @@ export type ErrorReason =
   | 'tool_execution_failed'
   | 'summary_after_cap_failed';
 
-// v1.8.2: shapes stored in messages.metadata. Discriminated on `kind`.
-//   cap_hit  — system sentinel emitted when tool budget is exhausted
-//   error    — attached to a failed assistant message so UI can show reason
+// v1.8.2 / v1.11.6: shapes stored in messages.metadata. Discriminated on `kind`.
+//   cap_hit    — system sentinel emitted when tool budget is exhausted
+//   doom_loop  — system sentinel emitted when the model called the same
+//                tool with the same args DOOM_LOOP_THRESHOLD times in a row
+//   error      — attached to a failed assistant message so UI can show reason
 export type MessageMetadata =
   | {
       kind: 'cap_hit';
@@ -139,6 +141,12 @@ export type MessageMetadata =
       agent_name: string | null;
       can_continue: boolean;
     }
+  | {
+      kind: 'doom_loop';
+      tool_name: string;
+      args: Record<string, unknown>;
+      threshold: number;
+    }
   | {
       kind: 'error';
       error_reason: ErrorReason;