unbrowse 3.7.0 → 3.7.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/cli.js +52 -17
- package/dist/mcp.js +4 -4
- package/dist/server.js +191 -34
- package/package.json +1 -1
package/dist/cli.js
CHANGED
|
@@ -31,7 +31,7 @@ var __promiseAll = (args) => Promise.all(args);
|
|
|
31
31
|
var __require = /* @__PURE__ */ createRequire(import.meta.url);
|
|
32
32
|
|
|
33
33
|
// ../../src/build-info.generated.ts
|
|
34
|
-
var BUILD_RELEASE_VERSION = "3.7.
|
|
34
|
+
var BUILD_RELEASE_VERSION = "3.7.1", BUILD_GIT_SHA = "90d1969f3abe", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiMy43LjEiLCJnaXRfc2hhIjoiOTBkMTk2OWYzYWJlIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUA5MGQxOTY5ZjNhYmUiLCJpc3N1ZWRfYXQiOiIyMDI2LTA0LTEwVDA4OjQyOjEwLjg1NloifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "R5-PEYUVNGWtRmubf9WAMDDuH1_C2N6D4xteuGelBIM", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
|
|
35
35
|
|
|
36
36
|
// ../../src/version.ts
|
|
37
37
|
import { createHash } from "crypto";
|
|
@@ -3375,43 +3375,59 @@ async function bootstrapAgentMailKey() {
|
|
|
3375
3375
|
let currentUrl = await getCurrentUrl(tabId);
|
|
3376
3376
|
let onDashboard = typeof currentUrl === "string" && currentUrl.includes("/dashboard");
|
|
3377
3377
|
if (!onDashboard) {
|
|
3378
|
-
log("bootstrap-agentmail", "not logged in — looking for
|
|
3379
|
-
const
|
|
3380
|
-
// Clerk
|
|
3378
|
+
log("bootstrap-agentmail", "not logged in — looking for social OAuth");
|
|
3379
|
+
const clickResult = await evaluate(tabId, `(() => {
|
|
3380
|
+
// Clerk's social button class pattern
|
|
3381
|
+
const socialBtns = document.querySelectorAll('.cl-socialButtonsBlockButton, [data-provider]');
|
|
3382
|
+
for (const btn of socialBtns) {
|
|
3383
|
+
const text = (btn.textContent || '').toLowerCase();
|
|
3384
|
+
const provider = btn.getAttribute('data-provider') || '';
|
|
3385
|
+
// Match any social provider — Google, GitHub, etc.
|
|
3386
|
+
if (text || provider) {
|
|
3387
|
+
btn.click();
|
|
3388
|
+
return 'clicked:' + (provider || text.slice(0, 20));
|
|
3389
|
+
}
|
|
3390
|
+
}
|
|
3391
|
+
// Fallback: look for any button with a known OAuth provider name
|
|
3381
3392
|
const btns = document.querySelectorAll('button, a');
|
|
3382
3393
|
for (const btn of btns) {
|
|
3383
3394
|
const text = (btn.textContent || '').toLowerCase();
|
|
3384
|
-
|
|
3385
|
-
if (text.includes('github') || ariaLabel.includes('github') ||
|
|
3386
|
-
btn.querySelector('img[alt*="github" i], svg[aria-label*="github" i]')) {
|
|
3395
|
+
if (text.includes('google') || text.includes('github') || text.includes('continue with')) {
|
|
3387
3396
|
btn.click();
|
|
3388
|
-
return 'clicked';
|
|
3397
|
+
return 'clicked:' + text.slice(0, 30);
|
|
3389
3398
|
}
|
|
3390
3399
|
}
|
|
3391
|
-
// Also check for social login containers
|
|
3392
|
-
const social = document.querySelector('.cl-socialButtonsBlockButton__github, [data-provider="github"]');
|
|
3393
|
-
if (social) { social.click(); return 'clicked'; }
|
|
3394
3400
|
return 'not_found';
|
|
3395
3401
|
})()`);
|
|
3396
|
-
|
|
3397
|
-
|
|
3402
|
+
const clickStr = typeof clickResult === "string" ? clickResult : String(clickResult);
|
|
3403
|
+
log("bootstrap-agentmail", `social button result: ${clickStr}`);
|
|
3404
|
+
if (clickStr.startsWith("clicked")) {
|
|
3405
|
+
log("bootstrap-agentmail", "clicked social OAuth — waiting for redirect");
|
|
3398
3406
|
const start2 = Date.now();
|
|
3399
3407
|
while (Date.now() - start2 < AUTH_TIMEOUT_MS) {
|
|
3400
3408
|
await new Promise((r) => setTimeout(r, 2000));
|
|
3401
3409
|
currentUrl = await getCurrentUrl(tabId);
|
|
3402
3410
|
if (typeof currentUrl === "string" && currentUrl.includes("/dashboard")) {
|
|
3403
3411
|
onDashboard = true;
|
|
3404
|
-
log("bootstrap-agentmail", "
|
|
3412
|
+
log("bootstrap-agentmail", "OAuth completed — on dashboard");
|
|
3405
3413
|
break;
|
|
3406
3414
|
}
|
|
3407
3415
|
}
|
|
3408
3416
|
}
|
|
3409
3417
|
if (!onDashboard) {
|
|
3410
3418
|
log("bootstrap-agentmail", "could not auto-login — manual setup needed");
|
|
3419
|
+
const pageInfo = await evaluate(tabId, `JSON.stringify({
|
|
3420
|
+
url: window.location.href,
|
|
3421
|
+
title: document.title,
|
|
3422
|
+
visible_buttons: Array.from(document.querySelectorAll('button,a'))
|
|
3423
|
+
.map(e => (e.textContent || '').trim().slice(0, 30))
|
|
3424
|
+
.filter(t => t)
|
|
3425
|
+
.slice(0, 10),
|
|
3426
|
+
})`).catch(() => "{}");
|
|
3411
3427
|
return {
|
|
3412
3428
|
success: false,
|
|
3413
3429
|
method: "manual",
|
|
3414
|
-
error:
|
|
3430
|
+
error: `Could not auto-login to AgentMail console. Page state: ${pageInfo}. Sign up at https://console.agentmail.to and create an API key manually, then: export AGENTMAIL_API_KEY=<key>`
|
|
3415
3431
|
};
|
|
3416
3432
|
}
|
|
3417
3433
|
}
|
|
@@ -6792,6 +6808,11 @@ async function cmdSetup(flags) {
|
|
|
6792
6808
|
properties: { command: "setup" }
|
|
6793
6809
|
});
|
|
6794
6810
|
info("Running setup checks");
|
|
6811
|
+
try {
|
|
6812
|
+
await ensureRegistered({ promptForEmail: false, exitOnFailure: false });
|
|
6813
|
+
} catch (err) {
|
|
6814
|
+
info(`[setup] background registration issue: ${err instanceof Error ? err.message : err}`);
|
|
6815
|
+
}
|
|
6795
6816
|
const report = await runSetup({
|
|
6796
6817
|
cwd: process.cwd(),
|
|
6797
6818
|
opencode: normalizeSetupScope(flags.opencode),
|
|
@@ -7603,8 +7624,22 @@ async function cmdLoginAuto(args, flags) {
|
|
|
7603
7624
|
}
|
|
7604
7625
|
})();
|
|
7605
7626
|
const { autonomousEmailLogin: autonomousEmailLogin3, isAgentMailAvailable: isAgentMailAvailable3 } = await init_agent_mail2().then(() => exports_agent_mail2);
|
|
7606
|
-
if (!isAgentMailAvailable3())
|
|
7607
|
-
|
|
7627
|
+
if (!isAgentMailAvailable3()) {
|
|
7628
|
+
info(`[login-auto] AGENTMAIL_API_KEY not set \u2014 attempting auto-bootstrap via console.agentmail.to`);
|
|
7629
|
+
try {
|
|
7630
|
+
const { bootstrapAgentMailKey: bootstrapAgentMailKey2 } = await init_bootstrap_agentmail().then(() => exports_bootstrap_agentmail);
|
|
7631
|
+
const bootstrap = await bootstrapAgentMailKey2();
|
|
7632
|
+
if (bootstrap.success && bootstrap.api_key) {
|
|
7633
|
+
process.env.AGENTMAIL_API_KEY = bootstrap.api_key;
|
|
7634
|
+
info(`[login-auto] bootstrapped AgentMail key via ${bootstrap.method ?? "browser"}`);
|
|
7635
|
+
} else {
|
|
7636
|
+
return die(`AGENTMAIL_API_KEY not set and bootstrap failed: ${bootstrap.error ?? "unknown"}. Get a key at https://agentmail.to and run: export AGENTMAIL_API_KEY=<key>`);
|
|
7637
|
+
}
|
|
7638
|
+
} catch (err) {
|
|
7639
|
+
const msg = err instanceof Error ? err.message : String(err);
|
|
7640
|
+
return die(`AGENTMAIL_API_KEY not set and bootstrap errored: ${msg}. Get a key at https://agentmail.to and run: export AGENTMAIL_API_KEY=<key>`);
|
|
7641
|
+
}
|
|
7642
|
+
}
|
|
7608
7643
|
info(`[login-auto] creating agent email for ${domain}...`);
|
|
7609
7644
|
const session = await autonomousEmailLogin3(domain);
|
|
7610
7645
|
info(`[login-auto] email: ${session.email}`);
|
package/dist/mcp.js
CHANGED
|
@@ -225,11 +225,11 @@ import { dirname, join, parse } from "path";
|
|
|
225
225
|
import { fileURLToPath as fileURLToPath2 } from "url";
|
|
226
226
|
|
|
227
227
|
// ../../src/build-info.generated.ts
|
|
228
|
-
var BUILD_RELEASE_VERSION = "3.7.
|
|
229
|
-
var BUILD_GIT_SHA = "
|
|
228
|
+
var BUILD_RELEASE_VERSION = "3.7.1";
|
|
229
|
+
var BUILD_GIT_SHA = "90d1969f3abe";
|
|
230
230
|
var BUILD_CODE_HASH = "5d9ebf619c61";
|
|
231
|
-
var BUILD_RELEASE_MANIFEST_BASE64 = "
|
|
232
|
-
var BUILD_RELEASE_MANIFEST_SIGNATURE = "
|
|
231
|
+
var BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiMy43LjEiLCJnaXRfc2hhIjoiOTBkMTk2OWYzYWJlIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUA5MGQxOTY5ZjNhYmUiLCJpc3N1ZWRfYXQiOiIyMDI2LTA0LTEwVDA4OjQyOjEwLjg1NloifQ";
|
|
232
|
+
var BUILD_RELEASE_MANIFEST_SIGNATURE = "R5-PEYUVNGWtRmubf9WAMDDuH1_C2N6D4xteuGelBIM";
|
|
233
233
|
var BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
|
|
234
234
|
|
|
235
235
|
// ../../src/version.ts
|
package/dist/server.js
CHANGED
|
@@ -1910,6 +1910,22 @@ function deriveTemplateParamsFromContextUrl(urlTemplate, contextUrl) {
|
|
|
1910
1910
|
const actualValue = actualUrl.searchParams.get(key);
|
|
1911
1911
|
if (actualValue != null && actualValue !== "") {
|
|
1912
1912
|
out[placeholder] = actualValue;
|
|
1913
|
+
continue;
|
|
1914
|
+
}
|
|
1915
|
+
const byPlaceholderName = actualUrl.searchParams.get(placeholder);
|
|
1916
|
+
if (byPlaceholderName != null && byPlaceholderName !== "") {
|
|
1917
|
+
out[placeholder] = byPlaceholderName;
|
|
1918
|
+
continue;
|
|
1919
|
+
}
|
|
1920
|
+
const lowerPlaceholder = placeholder.toLowerCase();
|
|
1921
|
+
if (/^(q|query|search|text|keyword|keywords|term|terms)$/.test(lowerPlaceholder)) {
|
|
1922
|
+
for (const alias of ["q", "query", "search", "text", "keyword", "keywords", "term"]) {
|
|
1923
|
+
const v = actualUrl.searchParams.get(alias);
|
|
1924
|
+
if (v != null && v !== "") {
|
|
1925
|
+
out[placeholder] = v;
|
|
1926
|
+
break;
|
|
1927
|
+
}
|
|
1928
|
+
}
|
|
1913
1929
|
}
|
|
1914
1930
|
}
|
|
1915
1931
|
return out;
|
|
@@ -3827,6 +3843,13 @@ __export(exports_reverse_engineer, {
|
|
|
3827
3843
|
extractAuthHeaders: () => extractAuthHeaders
|
|
3828
3844
|
});
|
|
3829
3845
|
import { nanoid as nanoid2 } from "nanoid";
|
|
3846
|
+
import { createHash } from "node:crypto";
|
|
3847
|
+
function stableEndpointId(method, urlTemplate) {
|
|
3848
|
+
if (!method || !urlTemplate)
|
|
3849
|
+
return nanoid2();
|
|
3850
|
+
const hash = createHash("sha256").update(`${method}:${urlTemplate}`).digest("base64url");
|
|
3851
|
+
return hash.slice(0, 21);
|
|
3852
|
+
}
|
|
3830
3853
|
function extractGraphQLOperationName(url, requestBody) {
|
|
3831
3854
|
const urlMatch = url.match(/\/graphql\/[^/]+\/(\w+)/);
|
|
3832
3855
|
if (urlMatch)
|
|
@@ -4447,10 +4470,11 @@ function extractEndpoints(requests, wsMessages, context) {
|
|
|
4447
4470
|
});
|
|
4448
4471
|
const csrfPlan = inferCsrfPlan(req, parsedRequestBody);
|
|
4449
4472
|
const endpointGraphqlOp = /graphql/i.test(req.url) ? extractGraphQLOperationName(req.url, req.request_body) : undefined;
|
|
4473
|
+
const computedUrlTemplate = qTemplateStr ? `${pathTemplate}${pathTemplate.includes("?") ? "&" : "?"}${qTemplateStr}` : pathTemplate;
|
|
4450
4474
|
let endpoint = {
|
|
4451
|
-
endpoint_id:
|
|
4475
|
+
endpoint_id: stableEndpointId(req.method, computedUrlTemplate),
|
|
4452
4476
|
method: req.method,
|
|
4453
|
-
url_template:
|
|
4477
|
+
url_template: computedUrlTemplate,
|
|
4454
4478
|
description: buildEndpointDescription(req, sampleRequest, sampleResponse),
|
|
4455
4479
|
headers_template: sanitizeHeaders(req.request_headers),
|
|
4456
4480
|
query: sanitizedQParams,
|
|
@@ -4541,7 +4565,7 @@ function extractEndpoints(requests, wsMessages, context) {
|
|
|
4541
4565
|
response_schema = inferSchema(jsonSamples);
|
|
4542
4566
|
}
|
|
4543
4567
|
const endpoint = {
|
|
4544
|
-
endpoint_id:
|
|
4568
|
+
endpoint_id: stableEndpointId("WS", wsUrl),
|
|
4545
4569
|
method: "WS",
|
|
4546
4570
|
url_template: wsUrl,
|
|
4547
4571
|
idempotency: "safe",
|
|
@@ -7096,10 +7120,10 @@ var init_capture = __esm(async () => {
|
|
|
7096
7120
|
});
|
|
7097
7121
|
|
|
7098
7122
|
// ../../src/build-info.generated.ts
|
|
7099
|
-
var BUILD_RELEASE_VERSION = "3.7.
|
|
7123
|
+
var BUILD_RELEASE_VERSION = "3.7.1", BUILD_GIT_SHA = "90d1969f3abe", BUILD_CODE_HASH = "5d9ebf619c61", BUILD_RELEASE_MANIFEST_BASE64 = "eyJzY2hlbWFfdmVyc2lvbiI6MSwicmVsZWFzZV92ZXJzaW9uIjoiMy43LjEiLCJnaXRfc2hhIjoiOTBkMTk2OWYzYWJlIiwiY29kZV9oYXNoIjoiNWQ5ZWJmNjE5YzYxIiwidHJhY2VfdmVyc2lvbiI6IjVkOWViZjYxOWM2MUA5MGQxOTY5ZjNhYmUiLCJpc3N1ZWRfYXQiOiIyMDI2LTA0LTEwVDA4OjQyOjEwLjg1NloifQ", BUILD_RELEASE_MANIFEST_SIGNATURE = "R5-PEYUVNGWtRmubf9WAMDDuH1_C2N6D4xteuGelBIM", BUILD_DEFAULT_BACKEND_URL = "https://beta-api.unbrowse.ai";
|
|
7100
7124
|
|
|
7101
7125
|
// ../../src/version.ts
|
|
7102
|
-
import { createHash } from "crypto";
|
|
7126
|
+
import { createHash as createHash2 } from "crypto";
|
|
7103
7127
|
import { existsSync as existsSync4, readFileSync as readFileSync2, readdirSync } from "fs";
|
|
7104
7128
|
import { dirname, join as join3, parse } from "path";
|
|
7105
7129
|
import { fileURLToPath as fileURLToPath2 } from "url";
|
|
@@ -7116,7 +7140,7 @@ function collectTsFiles(dir) {
|
|
|
7116
7140
|
return results;
|
|
7117
7141
|
}
|
|
7118
7142
|
function hashFiles(srcDir, files) {
|
|
7119
|
-
const hash =
|
|
7143
|
+
const hash = createHash2("sha256");
|
|
7120
7144
|
for (const file of files) {
|
|
7121
7145
|
hash.update(file.slice(srcDir.length));
|
|
7122
7146
|
hash.update(readFileSync2(file, "utf-8"));
|
|
@@ -7156,7 +7180,7 @@ function computeCodeHash() {
|
|
|
7156
7180
|
} catch {}
|
|
7157
7181
|
const pkgVersion = getPackageVersion();
|
|
7158
7182
|
if (pkgVersion !== "unknown") {
|
|
7159
|
-
return
|
|
7183
|
+
return createHash2("sha256").update(`package:${pkgVersion}`).digest("hex").slice(0, 12);
|
|
7160
7184
|
}
|
|
7161
7185
|
return "compiled";
|
|
7162
7186
|
}
|
|
@@ -7210,13 +7234,13 @@ var init_version = __esm(() => {
|
|
|
7210
7234
|
});
|
|
7211
7235
|
|
|
7212
7236
|
// ../../src/payments/cascade.ts
|
|
7213
|
-
import { createHash as
|
|
7237
|
+
import { createHash as createHash3 } from "crypto";
|
|
7214
7238
|
import bs58 from "bs58";
|
|
7215
7239
|
function payableContributors(skill) {
|
|
7216
7240
|
return (skill.contributors ?? []).filter((c) => !!c.wallet_address?.trim());
|
|
7217
7241
|
}
|
|
7218
7242
|
function cascadeLabel(skillId) {
|
|
7219
|
-
const digest =
|
|
7243
|
+
const digest = createHash3("sha256").update(skillId).digest("hex");
|
|
7220
7244
|
return `ubr-${digest.slice(0, 23)}`;
|
|
7221
7245
|
}
|
|
7222
7246
|
function decodeSecretKey(raw) {
|
|
@@ -7610,7 +7634,7 @@ __export(exports_client2, {
|
|
|
7610
7634
|
import { readFileSync as readFileSync4, writeFileSync as writeFileSync3, existsSync as existsSync7, mkdirSync as mkdirSync4, readdirSync as readdirSync2, unlinkSync } from "fs";
|
|
7611
7635
|
import { join as join6 } from "path";
|
|
7612
7636
|
import { homedir as homedir4, hostname, release as osRelease } from "os";
|
|
7613
|
-
import { randomBytes as randomBytes2, createHash as
|
|
7637
|
+
import { randomBytes as randomBytes2, createHash as createHash4 } from "crypto";
|
|
7614
7638
|
import { createInterface } from "readline";
|
|
7615
7639
|
import { execSync } from "child_process";
|
|
7616
7640
|
function buildReleaseAttestationHeaders(manifestBase64, signature) {
|
|
@@ -7887,7 +7911,7 @@ function getApiKey() {
|
|
|
7887
7911
|
function hashApiKey(key) {
|
|
7888
7912
|
if (!key || key === "local-only")
|
|
7889
7913
|
return "";
|
|
7890
|
-
return
|
|
7914
|
+
return createHash4("sha256").update(key).digest("hex");
|
|
7891
7915
|
}
|
|
7892
7916
|
function getAgentId() {
|
|
7893
7917
|
const config = loadConfig();
|
|
@@ -9069,7 +9093,7 @@ var init_publish_admission = __esm(() => {
|
|
|
9069
9093
|
});
|
|
9070
9094
|
|
|
9071
9095
|
// ../../src/telemetry.ts
|
|
9072
|
-
import { createHash as
|
|
9096
|
+
import { createHash as createHash5 } from "node:crypto";
|
|
9073
9097
|
import { existsSync as existsSync8, mkdirSync as mkdirSync5, writeFileSync as writeFileSync4 } from "node:fs";
|
|
9074
9098
|
import { join as join7 } from "node:path";
|
|
9075
9099
|
function getTraceDir() {
|
|
@@ -9079,7 +9103,7 @@ function isTracingEnabled() {
|
|
|
9079
9103
|
return process.env.UNBROWSE_DISABLE_TRACES !== "1";
|
|
9080
9104
|
}
|
|
9081
9105
|
function hashValue(value) {
|
|
9082
|
-
return
|
|
9106
|
+
return createHash5("sha256").update(value).digest("hex").slice(0, 16);
|
|
9083
9107
|
}
|
|
9084
9108
|
function isSensitiveName(name) {
|
|
9085
9109
|
return SENSITIVE_PATTERNS.some((re) => re.test(name));
|
|
@@ -9097,7 +9121,7 @@ function anonymizeUrl(url) {
|
|
|
9097
9121
|
}
|
|
9098
9122
|
function hashResponseBody(result) {
|
|
9099
9123
|
const str = typeof result === "string" ? result : JSON.stringify(result ?? "");
|
|
9100
|
-
return
|
|
9124
|
+
return createHash5("sha256").update(str).digest("hex").slice(0, 32);
|
|
9101
9125
|
}
|
|
9102
9126
|
function classifyFailure(error) {
|
|
9103
9127
|
if (!error)
|
|
@@ -12457,7 +12481,12 @@ function cleanDOM(html) {
|
|
|
12457
12481
|
const $ = cheerio.load(html);
|
|
12458
12482
|
for (const tag of STRIP_TAGS) {
|
|
12459
12483
|
if (tag === "script") {
|
|
12460
|
-
$("script").
|
|
12484
|
+
$("script").each((_, el) => {
|
|
12485
|
+
const type = $(el).attr("type") ?? "";
|
|
12486
|
+
if (type !== "application/ld+json") {
|
|
12487
|
+
$(el).remove();
|
|
12488
|
+
}
|
|
12489
|
+
});
|
|
12461
12490
|
} else {
|
|
12462
12491
|
$(tag).remove();
|
|
12463
12492
|
}
|
|
@@ -15495,7 +15524,12 @@ __export(exports_execution, {
|
|
|
15495
15524
|
buildCanonicalDocumentEndpoint: () => buildCanonicalDocumentEndpoint
|
|
15496
15525
|
});
|
|
15497
15526
|
import { nanoid as nanoid6 } from "nanoid";
|
|
15498
|
-
import { createHash as
|
|
15527
|
+
import { createHash as createHash6 } from "node:crypto";
|
|
15528
|
+
function stableEndpointId2(method, urlTemplate) {
|
|
15529
|
+
if (!method || !urlTemplate)
|
|
15530
|
+
return nanoid6();
|
|
15531
|
+
return createHash6("sha256").update(`${method}:${urlTemplate}`).digest("base64url").slice(0, 21);
|
|
15532
|
+
}
|
|
15499
15533
|
function stampTrace(trace) {
|
|
15500
15534
|
trace.trace_version = TRACE_VERSION;
|
|
15501
15535
|
return trace;
|
|
@@ -16039,10 +16073,11 @@ function buildPageArtifactCapture(url, intent, html, authRequired = false) {
|
|
|
16039
16073
|
const searchForms = detectSearchForms(html);
|
|
16040
16074
|
const validSearchForm = searchForms.find((spec) => isStructuredSearchForm(spec));
|
|
16041
16075
|
const response_schema = inferSchema([extracted.data]);
|
|
16076
|
+
const computedTemplate = templatizeQueryParams(url);
|
|
16042
16077
|
const endpoint = {
|
|
16043
|
-
endpoint_id:
|
|
16078
|
+
endpoint_id: stableEndpointId2("GET", computedTemplate),
|
|
16044
16079
|
method: "GET",
|
|
16045
|
-
url_template:
|
|
16080
|
+
url_template: computedTemplate,
|
|
16046
16081
|
idempotency: "safe",
|
|
16047
16082
|
verification_status: "verified",
|
|
16048
16083
|
reliability_score: extracted.confidence,
|
|
@@ -16090,7 +16125,7 @@ function buildCanonicalDocumentEndpoint(url, intent, authRequired = false) {
|
|
|
16090
16125
|
const replayTemplate = deriveStructuredDataReplayTemplate(url);
|
|
16091
16126
|
if (replayUrl === url && replayTemplate === url)
|
|
16092
16127
|
return;
|
|
16093
|
-
const canonicalId =
|
|
16128
|
+
const canonicalId = createHash6("sha1").update(replayTemplate !== url ? replayTemplate : replayUrl).digest("base64url").slice(0, 21);
|
|
16094
16129
|
const endpoint = {
|
|
16095
16130
|
endpoint_id: canonicalId,
|
|
16096
16131
|
method: "GET",
|
|
@@ -16270,6 +16305,65 @@ async function trySeedPublicDocumentFetchSkill(skill, url, intent, targetDomain,
|
|
|
16270
16305
|
redirect: "follow"
|
|
16271
16306
|
});
|
|
16272
16307
|
const html = await response.text();
|
|
16308
|
+
const contentType = (response.headers.get("content-type") || "").toLowerCase();
|
|
16309
|
+
if (response.ok && (contentType.includes("application/json") || contentType.includes("text/json"))) {
|
|
16310
|
+
try {
|
|
16311
|
+
const parsed = JSON.parse(html);
|
|
16312
|
+
const urlObj = new URL(response.url || url);
|
|
16313
|
+
const pathTemplate = `${urlObj.origin}${urlObj.pathname}`;
|
|
16314
|
+
const responseSchema = inferSchema([parsed]);
|
|
16315
|
+
const endpoint = {
|
|
16316
|
+
endpoint_id: stableEndpointId2("GET", pathTemplate),
|
|
16317
|
+
method: "GET",
|
|
16318
|
+
url_template: pathTemplate,
|
|
16319
|
+
idempotency: "safe",
|
|
16320
|
+
verification_status: "verified",
|
|
16321
|
+
reliability_score: 0.95,
|
|
16322
|
+
description: `Direct JSON API for ${intent}`,
|
|
16323
|
+
response_schema: responseSchema
|
|
16324
|
+
};
|
|
16325
|
+
endpoint.semantic = inferEndpointSemantic(endpoint, {
|
|
16326
|
+
sampleResponse: parsed,
|
|
16327
|
+
observedAt: new Date().toISOString(),
|
|
16328
|
+
sampleRequestUrl: url
|
|
16329
|
+
});
|
|
16330
|
+
const domain2 = getRegistrableDomain(targetDomain);
|
|
16331
|
+
const existingSkill2 = findExistingSkillForDomain(domain2, intent);
|
|
16332
|
+
const localEndpoints2 = await prepareLearnedEndpoints(existingSkill2 ? mergeEndpoints(existingSkill2.endpoints, [endpoint]) : [endpoint], intent, domain2);
|
|
16333
|
+
const localDraft2 = {
|
|
16334
|
+
skill_id: existingSkill2?.skill_id ?? nanoid6(),
|
|
16335
|
+
version: "1.0.0",
|
|
16336
|
+
schema_version: "1",
|
|
16337
|
+
lifecycle: "active",
|
|
16338
|
+
execution_type: "http",
|
|
16339
|
+
created_at: existingSkill2?.created_at ?? new Date().toISOString(),
|
|
16340
|
+
updated_at: new Date().toISOString(),
|
|
16341
|
+
name: domain2,
|
|
16342
|
+
intent_signature: intent,
|
|
16343
|
+
domain: domain2,
|
|
16344
|
+
description: `API skill for ${domain2}`,
|
|
16345
|
+
owner_type: "agent",
|
|
16346
|
+
endpoints: localEndpoints2,
|
|
16347
|
+
operation_graph: buildSkillOperationGraph(localEndpoints2),
|
|
16348
|
+
intents: [intent]
|
|
16349
|
+
};
|
|
16350
|
+
try {
|
|
16351
|
+
cachePublishedSkill(localDraft2);
|
|
16352
|
+
} catch {}
|
|
16353
|
+
return {
|
|
16354
|
+
trace: stampTrace({
|
|
16355
|
+
trace_id: nanoid6(),
|
|
16356
|
+
skill_id: localDraft2.skill_id,
|
|
16357
|
+
endpoint_id: endpoint.endpoint_id,
|
|
16358
|
+
started_at: new Date().toISOString(),
|
|
16359
|
+
completed_at: new Date().toISOString(),
|
|
16360
|
+
success: true,
|
|
16361
|
+
status_code: response.status
|
|
16362
|
+
}),
|
|
16363
|
+
result: parsed
|
|
16364
|
+
};
|
|
16365
|
+
} catch {}
|
|
16366
|
+
}
|
|
16273
16367
|
if (!isHtml(html) || isSpaShell(html))
|
|
16274
16368
|
return;
|
|
16275
16369
|
const built = buildPageArtifactCapture(response.url || url, intent, html, usedStoredAuth);
|
|
@@ -16595,7 +16689,7 @@ async function executeBrowserCapture(skill, params, options) {
|
|
|
16595
16689
|
epUrl = `${route.url}?${qStr}`;
|
|
16596
16690
|
}
|
|
16597
16691
|
endpoints.push({
|
|
16598
|
-
endpoint_id:
|
|
16692
|
+
endpoint_id: stableEndpointId2("GET", epUrl),
|
|
16599
16693
|
method: "GET",
|
|
16600
16694
|
url_template: epUrl,
|
|
16601
16695
|
query: epQuery,
|
|
@@ -19117,10 +19211,10 @@ var init_timing_economics = __esm(() => {
|
|
|
19117
19211
|
});
|
|
19118
19212
|
|
|
19119
19213
|
// ../../src/routing-telemetry.ts
|
|
19120
|
-
import { createHash as
|
|
19214
|
+
import { createHash as createHash7 } from "node:crypto";
|
|
19121
19215
|
import { nanoid as nanoid8 } from "nanoid";
|
|
19122
19216
|
function stableHash(value) {
|
|
19123
|
-
return
|
|
19217
|
+
return createHash7("sha256").update(JSON.stringify(value)).digest("hex").slice(0, 24);
|
|
19124
19218
|
}
|
|
19125
19219
|
function sanitizeScalar(value) {
|
|
19126
19220
|
if (value == null)
|
|
@@ -19474,7 +19568,7 @@ function applyVerificationResults(skill, verification) {
|
|
|
19474
19568
|
import { nanoid as nanoid9 } from "nanoid";
|
|
19475
19569
|
import { existsSync as existsSync13, writeFileSync as writeFileSync8, readFileSync as readFileSync8, mkdirSync as mkdirSync9, readdirSync as readdirSync6 } from "node:fs";
|
|
19476
19570
|
import { dirname as dirname2, join as join12 } from "node:path";
|
|
19477
|
-
import { createHash as
|
|
19571
|
+
import { createHash as createHash8 } from "node:crypto";
|
|
19478
19572
|
function summarizeSchema(schema, maxDepth = 3) {
|
|
19479
19573
|
function walk(s, depth) {
|
|
19480
19574
|
if (depth <= 0)
|
|
@@ -19602,7 +19696,7 @@ function scopedResolveCacheKeys(scope, key) {
|
|
|
19602
19696
|
return scope === "global" ? [scopedCacheKey("global", key)] : [scopedCacheKey(scope, key), scopedCacheKey("global", key)];
|
|
19603
19697
|
}
|
|
19604
19698
|
function snapshotPathForCacheKey(cacheKey) {
|
|
19605
|
-
const digest =
|
|
19699
|
+
const digest = createHash8("sha1").update(cacheKey).digest("hex");
|
|
19606
19700
|
return join12(SKILL_SNAPSHOT_DIR, `${digest}.json`);
|
|
19607
19701
|
}
|
|
19608
19702
|
function writeSkillSnapshot(cacheKey, skill) {
|
|
@@ -19953,7 +20047,7 @@ function buildLocalCanonicalReplaySkill(intent, contextUrl) {
|
|
|
19953
20047
|
const domain = new URL(contextUrl).hostname.replace(/^www\./, "");
|
|
19954
20048
|
const now = new Date().toISOString();
|
|
19955
20049
|
const skill = {
|
|
19956
|
-
skill_id: `canonical-${
|
|
20050
|
+
skill_id: `canonical-${createHash8("sha1").update(contextUrl).digest("hex").slice(0, 12)}`,
|
|
19957
20051
|
version: "1.0.0",
|
|
19958
20052
|
schema_version: "1",
|
|
19959
20053
|
name: `Canonical replay for ${domain}`,
|
|
@@ -22237,15 +22331,15 @@ async function resolveAndExecute(intent, params = {}, context, projection, optio
|
|
|
22237
22331
|
}
|
|
22238
22332
|
}
|
|
22239
22333
|
}
|
|
22240
|
-
if (context?.url
|
|
22334
|
+
if (context?.url) {
|
|
22241
22335
|
try {
|
|
22242
22336
|
const directRes = await fetch(context.url, {
|
|
22243
|
-
headers: { Accept: "application/json", "User-Agent": "unbrowse/1.0" },
|
|
22244
|
-
signal: AbortSignal.timeout(
|
|
22337
|
+
headers: { Accept: "application/json, text/html;q=0.5", "User-Agent": "unbrowse/1.0" },
|
|
22338
|
+
signal: AbortSignal.timeout(15000),
|
|
22245
22339
|
redirect: "follow"
|
|
22246
22340
|
});
|
|
22247
22341
|
const ct = directRes.headers.get("content-type") ?? "";
|
|
22248
|
-
if (directRes.ok && (ct.includes("json") || ct.includes("+json"))) {
|
|
22342
|
+
if (directRes.ok && (ct.includes("application/json") || ct.includes("+json") || ct.includes("text/json"))) {
|
|
22249
22343
|
const data = await directRes.json();
|
|
22250
22344
|
const trace2 = {
|
|
22251
22345
|
trace_id: nanoid9(),
|
|
@@ -22265,7 +22359,10 @@ async function resolveAndExecute(intent, params = {}, context, projection, optio
|
|
|
22265
22359
|
timing: t
|
|
22266
22360
|
};
|
|
22267
22361
|
}
|
|
22268
|
-
} catch {
|
|
22362
|
+
} catch (err) {
|
|
22363
|
+
const msg = err instanceof Error ? err.message : String(err);
|
|
22364
|
+
console.log(`[direct-fetch] ${context.url} skipped: ${msg.slice(0, 100)}`);
|
|
22365
|
+
}
|
|
22269
22366
|
}
|
|
22270
22367
|
if (process.env.UNBROWSE_LOCAL_ONLY === "1" && !forceCapture) {
|
|
22271
22368
|
return buildNoCachedMatch();
|
|
@@ -22424,6 +22521,32 @@ async function resolveAndExecute(intent, params = {}, context, projection, optio
|
|
|
22424
22521
|
throw error;
|
|
22425
22522
|
}
|
|
22426
22523
|
timing.execute_ms = Date.now() - te0;
|
|
22524
|
+
const captureErrCheck = result?.error;
|
|
22525
|
+
if (captureErrCheck === "connection_failed" || captureErrCheck === "capture_failed") {
|
|
22526
|
+
console.warn(`[capture] ${captureErrCheck} detected — restarting Kuri and retrying once`);
|
|
22527
|
+
try {
|
|
22528
|
+
const kuri = await Promise.resolve().then(() => (init_client(), exports_client));
|
|
22529
|
+
await kuri.stop().catch(() => {});
|
|
22530
|
+
await new Promise((r) => setTimeout(r, 500));
|
|
22531
|
+
} catch {}
|
|
22532
|
+
try {
|
|
22533
|
+
const retryCaptureSkill = await getOrCreateBrowserCaptureSkill();
|
|
22534
|
+
const retryOut = await withAbortableOpTimeout("live_capture_retry", LIVE_CAPTURE_TIMEOUT_MS, (signal) => executeSkill(retryCaptureSkill, { ...params, url: context.url, intent }, undefined, {
|
|
22535
|
+
...options,
|
|
22536
|
+
intent,
|
|
22537
|
+
contextUrl: context?.url,
|
|
22538
|
+
signal
|
|
22539
|
+
}));
|
|
22540
|
+
if (retryOut.trace.success || !retryOut.result?.error) {
|
|
22541
|
+
trace = retryOut.trace;
|
|
22542
|
+
result = retryOut.result;
|
|
22543
|
+
learned_skill = retryOut.learned_skill;
|
|
22544
|
+
console.log(`[capture] retry after Kuri restart succeeded`);
|
|
22545
|
+
}
|
|
22546
|
+
} catch (retryErr) {
|
|
22547
|
+
console.warn(`[capture] retry failed: ${retryErr instanceof Error ? retryErr.message : retryErr}`);
|
|
22548
|
+
}
|
|
22549
|
+
}
|
|
22427
22550
|
const captureResult = result;
|
|
22428
22551
|
const authRecommended = captureResult?.auth_recommended === true;
|
|
22429
22552
|
const directDomCaptureResult = trace.success && trace.endpoint_id !== "browser-capture" && !!result && typeof result === "object" && "_extraction" in result;
|
|
@@ -24207,7 +24330,13 @@ var init_browse_session = __esm(() => {
|
|
|
24207
24330
|
|
|
24208
24331
|
// ../../src/api/browse-index.ts
|
|
24209
24332
|
import { nanoid as nanoid10 } from "nanoid";
|
|
24333
|
+
import { createHash as createHash9 } from "node:crypto";
|
|
24210
24334
|
import { readFileSync as readFileSync12 } from "node:fs";
|
|
24335
|
+
function stableEndpointId3(method, urlTemplate) {
|
|
24336
|
+
if (!method || !urlTemplate)
|
|
24337
|
+
return nanoid10();
|
|
24338
|
+
return createHash9("sha256").update(`${method}:${urlTemplate}`).digest("base64url").slice(0, 21);
|
|
24339
|
+
}
|
|
24211
24340
|
function normalizeBrowseUrl(url, baseUrl) {
|
|
24212
24341
|
if (!url)
|
|
24213
24342
|
return url;
|
|
@@ -24385,7 +24514,7 @@ async function cacheBrowseRequests(params) {
|
|
|
24385
24514
|
return { domain, indexed: false, mode: "none", skill: null };
|
|
24386
24515
|
const urlTemplate = templatizeQueryParams2(sessionUrl);
|
|
24387
24516
|
const endpoint = {
|
|
24388
|
-
endpoint_id:
|
|
24517
|
+
endpoint_id: stableEndpointId3("GET", urlTemplate),
|
|
24389
24518
|
method: "GET",
|
|
24390
24519
|
url_template: urlTemplate,
|
|
24391
24520
|
idempotency: "safe",
|
|
@@ -26437,11 +26566,18 @@ async function registerRoutes(app) {
|
|
|
26437
26566
|
app.addHook("onRequest", async (req, reply) => {
|
|
26438
26567
|
if (req.url === "/health" || req.url === "/v1/stats" || req.url.startsWith("/v1/settings"))
|
|
26439
26568
|
return;
|
|
26440
|
-
|
|
26569
|
+
let key = getApiKey();
|
|
26570
|
+
if (!key) {
|
|
26571
|
+
try {
|
|
26572
|
+
const { waitForBackgroundRegistration: waitForBackgroundRegistration2 } = await Promise.resolve().then(() => (init_client2(), exports_client2));
|
|
26573
|
+
await waitForBackgroundRegistration2(1e4);
|
|
26574
|
+
key = getApiKey();
|
|
26575
|
+
} catch {}
|
|
26576
|
+
}
|
|
26441
26577
|
if (!key) {
|
|
26442
26578
|
return reply.code(401).send({
|
|
26443
26579
|
error: "api_key_required",
|
|
26444
|
-
message: "No API key configured.
|
|
26580
|
+
message: "No API key configured. Run `unbrowse setup` to register, or set UNBROWSE_API_KEY manually.",
|
|
26445
26581
|
docs_url: "https://unbrowse.ai"
|
|
26446
26582
|
});
|
|
26447
26583
|
}
|
|
@@ -26691,7 +26827,28 @@ async function registerRoutes(app) {
|
|
|
26691
26827
|
...context_url && typeof params?.url !== "string" ? { url: context_url } : {}
|
|
26692
26828
|
};
|
|
26693
26829
|
try {
|
|
26694
|
-
|
|
26830
|
+
let execResult = await executeSkill(skill, execParams, projection, { confirm_unsafe, confirm_third_party_terms, dry_run, skip_robots_check, intent, contextUrl: context_url, client_scope: clientScope });
|
|
26831
|
+
if (!execResult.trace.success && execResult.result?.error === "endpoint_not_found" && typeof execParams.endpoint_id === "string") {
|
|
26832
|
+
let recovered = false;
|
|
26833
|
+
const freshSkill = await getSkill2(skill_id, clientScope);
|
|
26834
|
+
if (freshSkill && freshSkill.endpoints.some((e) => e.endpoint_id === execParams.endpoint_id)) {
|
|
26835
|
+
console.log(`[exec] endpoint ${execParams.endpoint_id} found in fresh marketplace skill — retrying`);
|
|
26836
|
+
skill = freshSkill;
|
|
26837
|
+
recovered = true;
|
|
26838
|
+
}
|
|
26839
|
+
if (!recovered && context_url) {
|
|
26840
|
+
const { buildCanonicalDocumentEndpoint: buildCanonicalDocumentEndpoint2 } = await init_execution().then(() => exports_execution);
|
|
26841
|
+
const canonical = buildCanonicalDocumentEndpoint2(context_url, intent ?? "", false);
|
|
26842
|
+
if (canonical && canonical.endpoint_id === execParams.endpoint_id) {
|
|
26843
|
+
console.log(`[exec] endpoint ${execParams.endpoint_id} is a canonical replay — injecting into skill`);
|
|
26844
|
+
skill = { ...skill, endpoints: [canonical, ...skill.endpoints] };
|
|
26845
|
+
recovered = true;
|
|
26846
|
+
}
|
|
26847
|
+
}
|
|
26848
|
+
if (recovered) {
|
|
26849
|
+
execResult = await executeSkill(skill, execParams, projection, { confirm_unsafe, confirm_third_party_terms, dry_run, skip_robots_check, intent, contextUrl: context_url, client_scope: clientScope });
|
|
26850
|
+
}
|
|
26851
|
+
}
|
|
26695
26852
|
saveTrace(execResult.trace);
|
|
26696
26853
|
if (execResult.trace.endpoint_id) {
|
|
26697
26854
|
recordExecution(skill.skill_id, execResult.trace.endpoint_id, execResult.trace, skill).catch(() => {});
|