Self-hosted, Self-Improving, Federated Everything App — dev management, agent orchestration, persistent mind, digital twin, second brain, and creative pipeline from a single dashboard over Tailscale to port the user into the OS
JavaScript ai-agentsdashboarddeveloper-toolsdigital-twinquantified-selfsecond-brainself-hostedtailscale
import { describe, it, expect } from 'vitest';
import { extractFinalSummary, extractSimplifySummaries } from './agentSummaryExtraction.js';
import { LIFECYCLE_LINES } from '../lib/agentOutputMarkers.js';
describe('sentinel summary precedence', () => {
it.each(['Thinking: I will now run /simplify.', 'tokens used\n123\nEarlier assistant reply.'])(
'renders the latest completion summary instead of transcript content: %s', (transcript) => {
const summary = '## Summary\nFixed the reported bug.\n\nDocumented `✅ Agent signaled completion` behavior.\n✅ Tests passed';
const output = [
'✅ Agent signaled completion',
'An earlier completion attempt.',
transcript,
'✅ Agent signaled completion',
summary,
LIFECYCLE_LINES.shellTabHint(),
].join('\n');
expect(extractFinalSummary(output)).toBe(summary);
expect(extractSimplifySummaries(output)).toEqual({ taskSummary: summary, simplifySummary: null });
}
);
});
describe('extractFinalSummary', () => {
it('returns the last block of non-tool text', () => {
const output = [
'Let me read the code.',
'🔧 Using Read...',
' → …/services/cos.js',
'Found the bug. Here is the fix:',
'- Fixed null check on line 42',
'- Added error handling',
].join('\n');
expect(extractFinalSummary(output)).toBe(
'Found the bug. Here is the fix:\n- Fixed null check on line 42\n- Added error handling'
);
});
it('returns null for empty output', () => {
expect(extractFinalSummary('')).toBeNull();
expect(extractFinalSummary(null)).toBeNull();
});
it('extracts only the assistant tail from Codex output, ignoring earlier diff/exec dumps', () => {
const output = [
'server/services/imageGen/codex.test.js:300: // diff',
'server/services/imageGen/codex.test.js:307: // more diff',
'exec',
'codex',
'I am about to make changes.',
'apply patch',
'patch: completed',
'diff --git a/foo.js b/foo.js',
'@@ -1 +1 @@',
'-old',
'+new',
'tokens used',
'285,345',
'Implemented the TUI provider path.',
'- New `type: "tui"` provider support.',
'- New PTY-backed runner.',
].join('\n');
expect(extractFinalSummary(output)).toBe(
'Implemented the TUI provider path.\n- New `type: "tui"` provider support.\n- New PTY-backed runner.'
);
});
it('handles legacy Codex inline "tokens used: <n>" format', () => {
const output = [
'codex',
'doing stuff',
'apply patch',
'+added',
'tokens used: 12345',
'Final assistant reply line 1.',
'Final assistant reply line 2.',
].join('\n');
expect(extractFinalSummary(output)).toBe(
'Final assistant reply line 1.\nFinal assistant reply line 2.'
);
});
});
describe('extractSimplifySummaries', () => {
it('splits output at the /simplify boundary', () => {
const output = [
'Let me investigate.',
'🔧 Using Read...',
' → …/services/cos.js',
'Fixed the bug. Here is what I did:',
'- Fixed null check',
'- Added validation',
'Now let me run `/simplify` as instructed.',
'🔧 Using Skill...',
' → simplify',
'All three reviews confirm the code is clean.',
'- No DRY violations',
'- No issues found',
].join('\n');
const result = extractSimplifySummaries(output);
expect(result).not.toBeNull();
expect(result.taskSummary).toBe(
'Fixed the bug. Here is what I did:\n- Fixed null check\n- Added validation'
);
expect(result.simplifySummary).toBe(
'All three reviews confirm the code is clean.\n- No DRY violations\n- No issues found'
);
});
it('returns null when no /simplify marker is found', () => {
const output = 'Just a regular agent output.\nNo simplify here.';
expect(extractSimplifySummaries(output)).toBeNull();
});
it('returns null for empty input', () => {
expect(extractSimplifySummaries('')).toBeNull();
expect(extractSimplifySummaries(null)).toBeNull();
});
it('handles /simplify marker at the beginning with no pre-summary', () => {
const output = [
'Now running `/simplify` review.',
'🔧 Using Read...',
'Code is clean.',
].join('\n');
const result = extractSimplifySummaries(output);
// No task summary before the marker
expect(result.taskSummary).toBeNull();
expect(result.simplifySummary).toBe('Code is clean.');
});
it('treats Codex output as a single task summary (no /simplify split, even if the diff quotes /simplify)', () => {
// Mirrors the agent-5f6951e3 regression: Codex dumps source code that
// contains both "/simplify" and a "run" verb on the same line. The old
// logic split there and serialized 9000+ lines of grep output as the
// task summary.
const output = [
'server/services/cos.js:1596:9. Run `/simplify` to review changed code.',
'client/src/components/cos/tabs/AgentCard.jsx:697: title="Will run /simplify before committing">',
'exec',
'apply patch',
'diff --git a/foo b/foo',
'+something',
'tokens used',
'12000',
'Implemented the requested change.',
'- Did X.',
'- Did Y.',
].join('\n');
const result = extractSimplifySummaries(output);
expect(result).toEqual({
taskSummary: 'Implemented the requested change.\n- Did X.\n- Did Y.',
simplifySummary: null
});
});
it('matches various /simplify narration patterns', () => {
const patterns = [
'Now let me run `/simplify` as instructed.',
'Now running `/simplify` as required by the instructions:',
'Let me now launch /simplify to review.',
];
for (const line of patterns) {
const output = `Task done.\n${line}\nClean code.`;
const result = extractSimplifySummaries(output);
expect(result).not.toBeNull();
expect(result.taskSummary).toBe('Task done.');
expect(result.simplifySummary).toBe('Clean code.');
}
});
});