mixdog 0.7.18 → 0.8.1

package/scripts/boot-smoke.mjs

@@ -0,0 +1,94 @@
+#!/usr/bin/env node
+import { spawnSync } from 'node:child_process';
+import { dirname, resolve } from 'node:path';
+import { fileURLToPath } from 'node:url';
+
+const root = resolve(dirname(fileURLToPath(import.meta.url)), '..');
+const THRESHOLD_MS = Number(process.env.MIXDOG_BOOT_SMOKE_LIMIT_MS || 5_000);
+const FAST_BACKGROUND_ENV = {
+  MIXDOG_DISABLE_SESSION_PREWARM: '1',
+  MIXDOG_DISABLE_PROVIDER_WARMUP: '1',
+  MIXDOG_DISABLE_CHANNEL_START: '1',
+};
+
+function runCase(name, args, { env = {}, input = null, expectStdout = null, expectStderr = null } = {}) {
+  const startedAt = performance.now();
+  const result = spawnSync(process.execPath, args, {
+    cwd: root,
+    env: { ...process.env, ...env },
+    input,
+    encoding: 'utf8',
+    stdio: [input == null ? 'ignore' : 'pipe', 'pipe', 'pipe'],
+    timeout: Math.max(10_000, THRESHOLD_MS + 5_000),
+  });
+  const ms = performance.now() - startedAt;
+  const stdout = result.stdout || '';
+  const stderr = result.stderr || '';
+  if (result.status !== 0) {
+    throw new Error(`${name} exited ${result.status}:\n${stderr || stdout}`);
+  }
+  if (expectStdout && !stdout.includes(expectStdout)) {
+    throw new Error(`${name} stdout missing ${expectStdout}:\n${stdout.slice(0, 1000)}`);
+  }
+  if (expectStderr && !stderr.includes(expectStderr)) {
+    throw new Error(`${name} stderr missing ${expectStderr}:\n${stderr.slice(0, 1000)}`);
+  }
+  if (ms > THRESHOLD_MS) {
+    throw new Error(`${name} boot smoke exceeded ${THRESHOLD_MS}ms (${ms.toFixed(1)}ms)`);
+  }
+  return { name, ms: Math.round(ms * 10) / 10 };
+}
+
+const rows = [
+  runCase('help', ['src/cli.mjs', '--help'], {
+    expectStdout: 'standalone mixdog CLI/TUI coding agent',
+  }),
+  runCase('tui_import', ['--input-type=module', '-e', `
+    const mod = await import('./src/tui/dist/index.mjs');
+    if (typeof mod.runTui !== 'function') throw new Error('runTui export missing');
+  `]),
+  runCase('runtime_tools', ['--input-type=module', '-e', `
+    const mod = await import('./src/mixdog-session-runtime.mjs');
+    if (typeof mod.createMixdogSessionRuntime !== 'function') throw new Error('runtime export missing');
+    const runtime = await mod.createMixdogSessionRuntime({ toolMode: 'full' });
+    try {
+      const status = runtime.toolsStatus();
+      const active = new Set(status.activeTools || []);
+      for (const name of ['read','code_graph','grep','glob','list','apply_patch','explore','bridge','tool_search','shell']) {
+        if (!active.has(name)) throw new Error('missing ' + name + ' in ' + [...active].join(','));
+      }
+      for (const name of ['bash','task']) {
+        if (active.has(name)) throw new Error('unexpected ' + name + ' in ' + [...active].join(','));
+      }
+      const result = runtime.selectTools('shell');
+      if (!result.already.includes('shell') || !result.added.includes('task')) {
+        throw new Error('shell alias should keep shell active and add task: ' + JSON.stringify(result));
+      }
+      const nextActive = new Set(result.status.activeTools || []);
+      for (const name of ['shell','task']) {
+        if (!nextActive.has(name)) throw new Error('selected tool missing ' + name + ' in ' + [...nextActive].join(','));
+      }
+      console.log('runtime_tools active=' + status.activeCount + '/' + status.count + ' selected=' + result.added.join(','));
+    } finally {
+      await runtime.close('runtime-tools-smoke', { waitForExit: false });
+    }
+  `], {
+    env: FAST_BACKGROUND_ENV,
+    expectStdout: 'runtime_tools active=',
+  }),
+  runCase('boot_profile', ['src/cli.mjs', '--help'], {
+    env: { MIXDOG_BOOT_PROFILE: '1' },
+    expectStdout: 'standalone mixdog CLI/TUI coding agent',
+    expectStderr: 'app:run:start',
+  }),
+  runCase('plain_quit', ['src/cli.mjs', '--plain'], {
+    env: FAST_BACKGROUND_ENV,
+    input: '/quit\n',
+    expectStdout: 'bye.',
+  }),
+];
+
+for (const row of rows) {
+  process.stdout.write(`${row.name}: ${row.ms}ms\n`);
+}
+process.stdout.write('boot smoke passed\n');

package/scripts/build-tui.mjs

@@ -0,0 +1,52 @@
+#!/usr/bin/env node
+/**
+ * scripts/build-tui.mjs — bundle the React/ink TUI (JSX) to plain ESM.
+ *
+ * mixdog is otherwise zero-build (plain .mjs), but the React/ink render
+ * layer needs JSX transpilation. esbuild compiles src/tui/*.jsx into a single
+ * ESM bundle at src/tui/dist/index.mjs.
+ *
+ * What is bundled vs external:
+ *   - bundled: our JSX only.
+ *   - external: React and Mixdog's checked-in Ink renderer.
+ *     Ink is loaded from vendor/ink directly, never by rewriting node_modules.
+ *
+ * Run:  node scripts/build-tui.mjs   (or `npm run build:tui`)
+ */
+import { build } from 'esbuild';
+import { dirname, join } from 'node:path';
+import { fileURLToPath } from 'node:url';
+
+const ROOT = join(dirname(fileURLToPath(import.meta.url)), '..');
+const SRC = join(ROOT, 'src', 'tui');
+const DIST_TO_VENDOR_INK = '../../../vendor/ink/build/index.js';
+
+const mixdogInkAliasPlugin = {
+  name: 'mixdog-ink-alias',
+  setup(build) {
+    build.onResolve({ filter: /^ink$/ }, () => ({
+      path: DIST_TO_VENDOR_INK,
+      external: true,
+    }));
+  },
+};
+
+await build({
+  entryPoints: [join(SRC, 'index.jsx')],
+  outfile: join(SRC, 'dist', 'index.mjs'),
+  bundle: true,
+  format: 'esm',
+  platform: 'node',
+  target: 'node22',
+  jsx: 'automatic',
+  // Keep package imports external like the original CLI flow. Local shared
+  // helpers are bundled so relative paths stay valid from src/tui/dist/.
+  // Only `ink` is redirected to Mixdog's checked-in renderer instead of
+  // node_modules/ink.
+  packages: 'external',
+  external: ['../vendor/*', '../../vendor/*'],
+  plugins: [mixdogInkAliasPlugin],
+  logLevel: 'info',
+});
+
+process.stdout.write('build:tui ok → src/tui/dist/index.mjs\n');

package/scripts/compact-smoke.mjs

@@ -0,0 +1,199 @@
+#!/usr/bin/env node
+import { compactActiveTurn, compactMessages, semanticCompactMessages, SUMMARY_PREFIX } from '../src/runtime/agent/orchestrator/session/compact.mjs';
+import { agentLoop } from '../src/runtime/agent/orchestrator/session/loop.mjs';
+import { estimateMessagesTokens } from '../src/runtime/agent/orchestrator/session/context-utils.mjs';
+import { autoCompactWindowForRoute, summarizeGatewayUsage } from '../src/vendor/statusline/src/gateway/route-meta.mjs';
+
+function assert(condition, message) {
+  if (!condition) throw new Error(message);
+}
+
+function textOf(messages) {
+  return messages.map((m) => (typeof m?.content === 'string' ? m.content : JSON.stringify(m?.content ?? ''))).join('\n---\n');
+}
+
+function findSummary(messages) {
+  return messages.find((m) => m?.role === 'user' && typeof m.content === 'string' && m.content.startsWith(SUMMARY_PREFIX));
+}
+
+const longBlock = 'analysis detail '.repeat(260);
+const oldHistory = [];
+for (let i = 0; i < 14; i += 1) {
+  oldHistory.push({ role: 'user', content: `old request ${i}: inspect src/runtime/agent/orchestrator/session/compact.mjs\n${longBlock}` });
+  oldHistory.push({
+    role: 'assistant',
+    content: i % 2 === 0 ? `noted invariant ${i}: preserve exact file paths` : '',
+    toolCalls: [{
+      id: `call_old_${i}`,
+      name: 'grep',
+      arguments: {
+        path: 'src/runtime/agent/orchestrator/session/compact.mjs',
+        pattern: 'compactMessages',
+        token: 'sk-live-secret-must-not-survive',
+      },
+    }],
+  });
+  oldHistory.push({ role: 'tool', toolCallId: `call_old_${i}`, content: `grep result ${i}: compactMessages at line ${650 + i}\n${'tool output '.repeat(220)}` });
+}
+
+const currentTurn = [
+  { role: 'user', content: 'current task: continue cleanup and quality check compact logic' },
+  {
+    role: 'assistant',
+    content: 'checking the latest path',
+    toolCalls: [{ id: 'call_latest', name: 'read', arguments: { path: 'src/runtime/agent/orchestrator/session/manager.mjs', line: 2438 } }],
+  },
+  { role: 'tool', toolCallId: 'call_latest', content: 'latest result must remain verbatim' },
+];
+
+const messages = [
+  { role: 'system', content: 'system rules stay mandatory' },
+  ...oldHistory,
+  ...currentTurn,
+];
+
+const beforeTokens = estimateMessagesTokens(messages);
+const compacted = compactMessages(messages, 2_400, { force: true, reserveTokens: 256 });
+const afterTokens = estimateMessagesTokens(compacted);
+const summary = findSummary(compacted);
+const compactedText = textOf(compacted);
+
+assert(summary, 'deterministic compact must insert an anchored summary');
+assert(afterTokens < beforeTokens, `compact should shrink token estimate: before=${beforeTokens} after=${afterTokens}`);
+assert(afterTokens <= 2_400 - 256, `compact must respect effective budget: after=${afterTokens}`);
+assert(compacted.some((m) => m?.content === 'current task: continue cleanup and quality check compact logic'), 'current user turn must be preserved');
+assert(compacted.some((m) => m?.toolCallId === 'call_latest' && m.content === 'latest result must remain verbatim'), 'latest tool result must be preserved');
+assert(summary.content.includes('src/runtime/agent/orchestrator/session/compact.mjs'), 'summary should preserve relevant file paths');
+assert(summary.content.includes('tool_calls=grep('), 'summary should preserve compact tool-call intent');
+assert(summary.content.includes('[redacted]'), 'summary should redact sensitive tool-call args');
+assert(!compactedText.includes('sk-live-secret-must-not-survive'), 'compacted transcript must not retain sensitive tool-call arg values');
+
+const claudeStyleBudget = 1_400;
+const claudeStyleCompacted = compactMessages(messages, claudeStyleBudget, { force: true, reserveTokens: 256 });
+const claudeStyleAfterTokens = estimateMessagesTokens(claudeStyleCompacted) + 256;
+assert(claudeStyleAfterTokens < beforeTokens * 0.6, `auto compact target should restart with headroom: before=${beforeTokens} after=${claudeStyleAfterTokens}`);
+assert(findSummary(claudeStyleCompacted), 'headroom compact must still insert a handoff summary');
+assert(claudeStyleCompacted.some((m) => m?.toolCallId === 'call_latest' && m.content === 'latest result must remain verbatim'), 'headroom compact must preserve latest tool result');
+
+const statusRoute = {
+  provider: 'openai',
+  model: 'gpt-5.5',
+  contextWindow: 950_000,
+  rawContextWindow: 1_000_000,
+  autoCompactTokenLimit: 950_000,
+};
+assert(autoCompactWindowForRoute(statusRoute) === 950_000, 'auto compact window should match the effective compact capacity');
+const usageSummary = summarizeGatewayUsage(statusRoute, { usage: { inputTokens: 900_000, outputTokens: 1 } });
+assert(usageSummary.contextUsedPct === 94.74, `statusline usage should use effective compact capacity: ${usageSummary.contextUsedPct}`);
+
+let deterministicFailed = false;
+try {
+  compactMessages([
+    { role: 'system', content: 'system rules stay mandatory' },
+    { role: 'user', content: 'single very large active turn' },
+    { role: 'assistant', toolCalls: [{ id: 'call_a', name: 'grep', arguments: { pattern: 'older' } }] },
+    { role: 'tool', toolCallId: 'call_a', content: 'older same-turn output '.repeat(1_400) },
+    { role: 'assistant', toolCalls: [{ id: 'call_b', name: 'read', arguments: { path: 'src/runtime/agent/orchestrator/session/compact.mjs' } }] },
+    { role: 'tool', toolCallId: 'call_b', content: 'latest same-turn output must remain' },
+  ], 1_100, { force: true });
+} catch {
+  deterministicFailed = true;
+}
+assert(deterministicFailed, 'deterministic compact should fail when there is no prior history before the active turn');
+
+const activeCompacted = compactActiveTurn([
+  { role: 'system', content: 'system rules stay mandatory' },
+  { role: 'user', content: 'single very large active turn' },
+  { role: 'assistant', content: 'older group', toolCalls: [{ id: 'call_a', name: 'grep', arguments: { pattern: 'older' } }] },
+  { role: 'tool', toolCallId: 'call_a', content: 'older same-turn output '.repeat(1_400) },
+  { role: 'assistant', content: 'latest group', toolCalls: [{ id: 'call_b', name: 'read', arguments: { path: 'src/runtime/agent/orchestrator/session/compact.mjs' } }] },
+  { role: 'tool', toolCallId: 'call_b', content: 'latest same-turn output must remain' },
+], 1_100, { minActiveTurnGroups: 1, maxToolOutputChars: 400 });
+const activeText = textOf(activeCompacted);
+assert(estimateMessagesTokens(activeCompacted) <= 1_100, 'active-turn compact should fit the budget');
+assert(activeCompacted.some((m) => m?.toolCallId === 'call_b' && m.content === 'latest same-turn output must remain'), 'active-turn compact must preserve latest tool result');
+assert(
+  activeText.includes('[mixdog pruned old tool output') || !activeCompacted.some((m) => m?.toolCallId === 'call_a'),
+  'active-turn compact should drop or prune older same-turn output',
+);
+assert(!activeText.includes('older same-turn output '.repeat(50).trim()), 'active-turn compact should remove the full older tool body');
+
+const activeSummaryCompacted = compactActiveTurn([
+  { role: 'system', content: 'system rules stay mandatory' },
+  { role: 'user', content: 'single active turn with older tool work to summarize' },
+  { role: 'assistant', content: 'older active-turn group', toolCalls: [{ id: 'call_active_old', name: 'grep', arguments: { pattern: 'older' } }] },
+  { role: 'tool', toolCallId: 'call_active_old', content: 'older same-turn output '.repeat(220) },
+  { role: 'assistant', content: 'latest group', toolCalls: [{ id: 'call_active_latest', name: 'read', arguments: { path: 'src/runtime/agent/orchestrator/session/compact.mjs' } }] },
+  { role: 'tool', toolCallId: 'call_active_latest', content: 'latest same-turn output must remain' },
+], 5_000, { minActiveTurnGroups: 1, maxToolOutputChars: 10_000, force: true });
+const activeSummary = findSummary(activeSummaryCompacted);
+assert(activeSummary, 'active-turn compact should summarize dropped older same-turn groups when the summary fits');
+assert(activeSummary.content.includes('tool_calls=grep('), 'active-turn summary should preserve older same-turn tool intent');
+assert(activeSummaryCompacted.some((m) => m?.toolCallId === 'call_active_latest' && m.content === 'latest same-turn output must remain'), 'active-turn summary path must preserve latest tool result');
+
+const forcedSingleGroup = compactActiveTurn([
+  { role: 'system', content: 'system rules stay mandatory' },
+  { role: 'user', content: 'single active turn with one huge latest tool group' },
+  { role: 'assistant', content: 'latest group', toolCalls: [{ id: 'call_single_latest', name: 'read', arguments: { path: 'src/runtime/agent/orchestrator/session/compact.mjs' } }] },
+  { role: 'tool', toolCallId: 'call_single_latest', content: 'latest same-turn output '.repeat(500) },
+], 5_000, { minActiveTurnGroups: 1, maxToolOutputChars: 400, force: true });
+assert(textOf(forcedSingleGroup).includes('[mixdog pruned old tool output'), 'forced active-turn compact should prune a single oversized group instead of returning unchanged');
+
+let semanticCalls = 0;
+const semanticProvider = {
+  name: 'semantic-smoke',
+  async send(_messages, _model, _tools, opts) {
+    semanticCalls += 1;
+    assert(opts?.maxOutputTokens === 4_096, 'semantic compact should request the summary output cap');
+    return { content: '## Goal\n- continue compact smoke\n\n## Constraints & Preferences\n- (none)\n\n## Progress\n### Done\n- older turn summarized\n\n### In Progress\n- (none)\n\n### Blocked\n- (none)\n\n## Key Decisions\n- (none)\n\n## Next Steps\n- continue\n\n## Critical Context\n- src/runtime/agent/orchestrator/session/compact.mjs\n\n## Relevant Files\n- src/runtime/agent/orchestrator/session/compact.mjs: compact logic' };
+  },
+};
+const semanticMessages = [
+  { role: 'system', content: 'system rules stay mandatory' },
+  { role: 'user', content: 'older request about src/runtime/agent/orchestrator/session/compact.mjs' },
+  { role: 'assistant', content: 'older answer with useful details' },
+  { role: 'user', content: 'current request should remain verbatim' },
+];
+const semanticNoop = await semanticCompactMessages(semanticProvider, semanticMessages, 'fake-model', 5_000, { tailTurns: 1 });
+assert(semanticNoop.semantic === false && semanticCalls === 0, 'semantic compact should still no-op below budget unless forced');
+const semanticForced = await semanticCompactMessages(semanticProvider, semanticMessages, 'fake-model', 5_000, { tailTurns: 1, force: true });
+assert(semanticForced.semantic === true && semanticCalls === 1, 'forced semantic compact should run even when the local estimate fits');
+assert(findSummary(semanticForced.messages), 'forced semantic compact should insert an anchored summary');
+
+const overflowRetryMessages = [{ role: 'system', content: 'system rules stay mandatory' }];
+let overflowIndex = 0;
+while (estimateMessagesTokens(overflowRetryMessages) + 512 < 8_800) {
+  overflowRetryMessages.push({ role: 'user', content: `older overflow retry request ${overflowIndex}: ${'important detail '.repeat(90)}` });
+  overflowRetryMessages.push({ role: 'assistant', content: `older overflow retry answer ${overflowIndex}: ${'implementation note '.repeat(90)}` });
+  overflowIndex += 1;
+}
+overflowRetryMessages.push({ role: 'user', content: 'current overflow retry task must stay verbatim' });
+const overflowInitialPressure = estimateMessagesTokens(overflowRetryMessages) + 512;
+assert(overflowInitialPressure < 10_800, `overflow retry fixture must stay below normal trigger: pressure=${overflowInitialPressure}`);
+assert(overflowInitialPressure > 7_200, `overflow retry fixture must exceed strict retry budget before compaction: pressure=${overflowInitialPressure}`);
+let overflowSendCount = 0;
+let overflowRetryPressure = 0;
+const overflowProvider = {
+  name: 'overflow-smoke',
+  async send(sentMessages) {
+    overflowSendCount += 1;
+    if (overflowSendCount === 1) throw new Error('input tokens exceeds the context window');
+    overflowRetryPressure = estimateMessagesTokens(sentMessages) + 512;
+    return { content: '<final-answer>overflow retry recovered</final-answer>', usage: { inputTokens: overflowRetryPressure, outputTokens: 1 } };
+  },
+};
+const overflowSession = {
+  id: 'compact-smoke-overflow',
+  owner: 'bridge',
+  provider: 'overflow-smoke',
+  model: 'fake-model',
+  contextWindow: 12_000,
+  rawContextWindow: 12_000,
+  compactBoundaryTokens: 12_000,
+  compaction: { auto: true, semantic: false },
+};
+await agentLoop(overflowProvider, overflowRetryMessages, 'fake-model', [], null, process.cwd(), { session: overflowSession, sessionId: overflowSession.id });
+assert(overflowSendCount === 2, `overflow retry should send exactly twice, sent=${overflowSendCount}`);
+assert(overflowRetryPressure <= 7_200, `overflow retry must use stricter 60% budget: pressure=${overflowRetryPressure}`);
+
+process.stdout.write('compact smoke passed ✓\n');