~95% on SimpleQA (e.g. Qwen3.6-27B on a 3090). Supports all local and cloud LLMs (llama.cpp, Ollama, Google, ...). 10+ search engines - arXiv, PubMed, your private documents. Everything Local & Encrypted.
Python academiaanthropicarxivbravedeep-researchencryptionhome-automationhomeserverlocallocal-deep-researchlocal-llmmistralollamaopenaipubmedresearchresearch-toolretrieval-augmented-generationsearxngself-hosted
/**
* XSS regression coverage for the context-overflow stats tables.
*
* The requested symbols are implemented in components/context-overflow.js,
* not components/progress.js. The production path is
* loadContextData() -> displayContextData() -> renderRequestsTable() /
* displayModelStats() / displayTruncatedRequests().
*
* Interpolation map (context-overflow.js):
* renderRequestsTable :439 timestamp is Date output; :434 research_id is
* URL-encoded + escaped; :441 model and :450 query use attribute-context
* escaping for titles and text-context escaping for display; :442 provider,
* :443 phase are escaped; :444-449 number cells use the local
* Number()/isFinite-coercing formatNumber; badges/classes are literals.
* displayModelStats :509 model title/display and :510 provider are escaped;
* :524/:528/:532/:540/:544/:548 are formatNumber-coerced; :514 truncation
* rate and :569 response time are finite-number-coerced. Computed widths/colors
* are numeric/hardcoded.
* displayTruncatedRequests :612 timestamp is Date output; :614 query and
* :616 model are escaped; :617-619 are formatNumber-coerced; :621 is URL-
* encoded. displayContextData's dynamic-section values are textContent or
* number-derived. This component has no cost/formatCurrency interpolation.
*/
import { beforeAll, beforeEach, describe, expect, it, vi } from 'vitest';
const MARKUP = `
<div id="loading"></div>
<div id="content"></div>
<div id="empty-no-data"></div>
<div id="empty-no-truncation"></div>
<div id="empty-no-context-data"></div>
<div id="warning-banner"></div>
<span id="warning-rate"></span>
<div id="context-overflow-section"></div>
<table><tbody id="requests-tbody"></tbody></table>
<span id="pagination-info"></span>
<button id="pagination-prev"></button>
<button id="pagination-next"></button>
`;
const MARKUP_PAYLOAD = '<img src=x onerror=alert(1)>';
const SVG_PAYLOAD = '<svg onload=alert(2)>';
const ATTRIBUTE_PAYLOAD = 'x" onmouseover="alert(3)';
beforeAll(async () => {
window.__VITEST_TEST__ = true;
globalThis.Chart = class { destroy() {} };
globalThis.URLValidator = { safeAssign: vi.fn() };
window.HTMLCanvasElement.prototype.getContext = () => null;
await import('@js/components/context-overflow.js');
});
beforeEach(() => {
// eslint-disable-next-line no-unsanitized/property -- hardcoded fixture markup.
document.body.innerHTML = MARKUP;
});
function payload({ allRequests = [], modelStats = [], modelTokenStats = [], truncated = [] } = {}) {
return {
status: 'success',
overview: {
truncation_rate: 5,
truncated_requests: 1,
requests_with_context_data: 1,
total_requests: 1,
avg_tokens_truncated: 20,
},
token_summary: { total_requests: 1 },
model_stats: modelStats,
model_token_stats: modelTokenStats,
recent_truncated: truncated,
chart_data: [],
context_limits: [],
phase_breakdown: [],
current_context_window: 8192,
all_requests: allRequests,
pagination: { page: 1, total_pages: 1 },
};
}
async function render(data) {
globalThis.fetch = vi.fn(() => Promise.resolve({
ok: true,
json: () => Promise.resolve(data),
}));
await window.contextOverflowController.loadContextData('30d');
}
function expectNoLiveMarkup() {
expect(document.querySelectorAll('img, script, svg, iframe, object, embed')).toHaveLength(0);
for (const element of document.querySelectorAll('*')) {
expect([...element.attributes].some(({ name }) => /^on/i.test(name))).toBe(false);
}
}
describe('context-overflow stats table XSS surfaces', () => {
it('renders hostile request strings inertly and coerces every request numeric cell', async () => {
await render(payload({
allRequests: [{
timestamp: '2026-01-02T03:04:05Z',
research_id: 'id?<svg>',
model: SVG_PAYLOAD,
provider: MARKUP_PAYLOAD,
research_phase: SVG_PAYLOAD,
research_query: MARKUP_PAYLOAD,
prompt_tokens: MARKUP_PAYLOAD,
completion_tokens: MARKUP_PAYLOAD,
total_tokens: MARKUP_PAYLOAD,
context_limit: MARKUP_PAYLOAD,
tokens_truncated: MARKUP_PAYLOAD,
context_truncated: true,
}],
}));
const row = document.querySelector('#requests-tbody tr');
expect(row).not.toBeNull();
expect(row.textContent).toContain(SVG_PAYLOAD);
expect(row.textContent).toContain(MARKUP_PAYLOAD);
expect(row.textContent).toContain('0');
expectNoLiveMarkup();
});
it('renders hostile truncated-request text inertly and coerces its numeric cells', async () => {
await render(payload({
truncated: [{
timestamp: '2026-01-02T03:04:05Z',
research_id: `id/${MARKUP_PAYLOAD}`,
research_query: MARKUP_PAYLOAD,
model: SVG_PAYLOAD,
prompt_tokens: MARKUP_PAYLOAD,
context_limit: MARKUP_PAYLOAD,
tokens_truncated: MARKUP_PAYLOAD,
}],
}));
const table = document.querySelector('#truncated-list table');
expect(table).not.toBeNull();
expect(table.textContent).toContain(MARKUP_PAYLOAD);
expect(table.textContent).toContain(SVG_PAYLOAD);
expect(table.textContent).toContain('0');
expectNoLiveMarkup();
});
it('keeps model-card names/providers and formatNumber-fed values inert', async () => {
await render(payload({
modelStats: [{
model: SVG_PAYLOAD,
provider: MARKUP_PAYLOAD,
total_requests: MARKUP_PAYLOAD,
truncated_count: MARKUP_PAYLOAD,
truncation_rate: 5,
avg_context_limit: MARKUP_PAYLOAD,
}],
modelTokenStats: [{
model: SVG_PAYLOAD,
provider: MARKUP_PAYLOAD,
min_prompt: MARKUP_PAYLOAD,
avg_prompt: MARKUP_PAYLOAD,
max_prompt: MARKUP_PAYLOAD,
}],
}));
const card = document.querySelector('#model-stats .ldr-model-card');
expect(card).not.toBeNull();
expect(card.textContent).toContain(SVG_PAYLOAD);
expect(card.textContent).toContain(MARKUP_PAYLOAD);
expect(card.textContent).toContain('0');
expectNoLiveMarkup();
});
// Regression: renderRequestsTable model titles must not allow quote breakouts.
it('prevents quote breakout in renderRequestsTable model title', async () => {
await render(payload({
allRequests: [{
timestamp: '2026-01-02T03:04:05Z', model: ATTRIBUTE_PAYLOAD,
provider: 'openai', prompt_tokens: 1, completion_tokens: 2,
total_tokens: 3, context_limit: 4, tokens_truncated: 0,
}],
}));
expectNoLiveMarkup();
});
// Regression: model-card truncation rates must be numerically coerced.
it('rejects raw model-card truncation-rate interpolation', async () => {
await render(payload({
modelStats: [{
model: 'safe-model', provider: 'openai', total_requests: 1,
truncated_count: 0, truncation_rate: MARKUP_PAYLOAD,
}],
}));
expectNoLiveMarkup();
});
// Regression: model-card response durations must be numerically coerced.
it('rejects raw model-card response-duration interpolation', async () => {
await render(payload({
modelStats: [{
model: 'safe-model', provider: 'openai', total_requests: 1,
truncated_count: 0, truncation_rate: 5,
}],
modelTokenStats: [{
model: 'safe-model', provider: 'openai', avg_response_time_ms: MARKUP_PAYLOAD,
}],
}));
expectNoLiveMarkup();
});
it('renders a benign end-to-end stats table with real numbers', async () => {
await render(payload({
allRequests: [{
timestamp: '2026-01-02T03:04:05Z', research_id: 'research-123',
model: 'gpt-4', provider: 'openai', research_phase: 'analysis',
research_query: 'quantum computing', prompt_tokens: 1234,
completion_tokens: 567, total_tokens: 1801, context_limit: 8192,
tokens_truncated: 42, context_truncated: true,
}],
modelStats: [{
model: 'gpt-4', provider: 'openai', total_requests: 12,
truncated_count: 2, truncation_rate: 16, avg_context_limit: 8192,
}],
modelTokenStats: [{
model: 'gpt-4', provider: 'openai', min_prompt: 100,
avg_prompt: 1200, max_prompt: 2000, avg_response_time_ms: 250,
}],
truncated: [{
timestamp: '2026-01-02T03:04:05Z', research_id: 'research-123',
research_query: 'quantum computing', model: 'gpt-4',
prompt_tokens: 1234, context_limit: 8192, tokens_truncated: 42,
}],
}));
expect(document.querySelectorAll('#requests-tbody tr')).toHaveLength(1);
expect(document.querySelector('#requests-tbody').textContent).toContain('1,234');
expect(document.querySelector('#requests-tbody').textContent).toContain('1,801');
expect(document.querySelector('#model-stats').textContent).toContain('1,200');
expect(document.querySelector('#truncated-list').textContent).toContain('8,192');
expectNoLiveMarkup();
});
});