dsh-browser-playwright-codex 0.7.8 → 0.8.1

package/lib/tool-types.d.ts

@@ -46,6 +46,16 @@ export interface ChallengeNode {
     readonly name?: string;
     /** Accessible role. Both built-in engines provide one; it gates the wording tier. */
     readonly role?: string;
+    /** Whether a checkbox/radio is checked; present on toggles. */
+    readonly checked?: boolean;
+    /** Whether an option is selected; present on options. */
+    readonly selected?: boolean;
+    /**
+     * Current value of a textbox/textarea/select. The no-progress fingerprint
+     * includes it so that filling a field registers as change — otherwise a
+     * normal fill looks like "no visible change" and trips the anti-bot guess.
+     */
+    readonly value?: string;
     readonly children?: readonly ChallengeNode[];
 }
 /** The slice of a snapshot the challenge detector reads. */

package/lib/tool.js

@@ -1,7 +1,7 @@
 import z from '@deepseek-ai/schemastery';
 import { defineTool } from '@deepseek-ai/dsh-tools';
 import { BlockAssembler, createUserMessage } from '@deepseek-ai/dsh-llm';
-import { BrowserError, cdpDisabled, evaluateDisabled } from "./errors.js";
+import { BrowserError, cdpDisabled, evaluateDisabled, normalizeEvaluateResult } from "./errors.js";
 import { registeredSuffixes } from "./contract.js";
 import { getEnabled, publishGates, publishLoginSites } from "./runtime-state.js";
 import { summarizeLoginSites } from "./login-sites.js";
@@ -138,13 +138,18 @@ export function apply(ctx, config) {
             waitFor: {
                 type: 'string',
                 enum: ['load', 'domcontentloaded', 'networkidle'],
-                description: 'Load condition to wait for. Defaults to load.',
+                description: 'Load condition to wait for. Defaults to domcontentloaded; load never fires on long-polling or ' +
+                    'download pages and hangs the call (issue #5).',
             },
         },
         output: { schema: SNAPSHOT_SCHEMA, render: renderSnapshotValue },
         async execute(args, exec) {
             const session = await acquire(exec);
-            const waitUntil = args.waitFor ?? 'load';
+            // domcontentloaded, not load: a page with an endless subresource (long
+            // poll, streaming download, ad frame that never settles) never reaches
+            // 'load', so the default burned the whole navigation timeout on pages
+            // that were perfectly usable already (issue #5).
+            const waitUntil = args.waitFor ?? 'domcontentloaded';
             return snapshotValue(await session.navigate(args.url, waitUntil, exec.signal));
         },
     }));
@@ -435,13 +440,18 @@ export function apply(ctx, config) {
             waitFor: {
                 type: 'string',
                 enum: ['load', 'domcontentloaded', 'networkidle'],
-                description: 'Load condition to wait for. Defaults to load.',
+                description: 'Load condition to wait for. Defaults to domcontentloaded; load never fires on long-polling or ' +
+                    'download pages and hangs the call (issue #5).',
             },
         },
         output: { schema: SNAPSHOT_SCHEMA, render: renderSnapshotValue },
         async execute(args, exec) {
             const session = await acquire(exec);
-            const waitUntil = args.waitFor ?? 'load';
+            // domcontentloaded, not load: a page with an endless subresource (long
+            // poll, streaming download, ad frame that never settles) never reaches
+            // 'load', so the default burned the whole navigation timeout on pages
+            // that were perfectly usable already (issue #5).
+            const waitUntil = args.waitFor ?? 'domcontentloaded';
             return snapshotValue(await session.openTab(args.url, waitUntil, exec.signal));
         },
     }));
@@ -553,6 +563,9 @@ export function apply(ctx, config) {
     registerMaybe('evaluate', defineTool({
         name: p('evaluate'),
         description: 'Evaluate one JavaScript expression in the page and return its JSON-serializable result. ' +
+            'Return value rules: an object, string, number, boolean or null comes back as-is; ' +
+            'a top-level ARRAY, undefined or a function is returned as a JSON string with a note saying it was converted ' +
+            '(wrap it yourself, e.g. {items: [...]} , if you need the array shape). ' +
             'Disabled by default; the deployment must enable allowEvaluate. ' +
             'IRON RULE: prefer read-only expressions to INSPECT state (value, outerHTML, checked, textContent) before ' +
             'concluding an interaction failed. If you write to the DOM directly (setting value/textContent), ' +
@@ -580,7 +593,8 @@ export function apply(ctx, config) {
                 throw evaluateDisabled();
             }
             const session = await acquire(exec);
-            return (await session.evaluate(args.expression, exec.signal));
+            const raw = await session.evaluate(args.expression, exec.signal);
+            return normalizeEvaluateResult(raw);
         },
     }));
     registerMaybe('cdp', defineTool({

package/lib/types.d.ts

@@ -21,6 +21,12 @@ export interface BrowserNode {
     readonly selected?: boolean;
     /** Whether the element is disabled. */
     readonly disabled?: boolean;
+    /**
+     * Current value of a textbox, textarea or select, clipped to the snapshot's
+     * text budget. Present so the model can see what a fill actually left behind
+     * and so the no-progress fingerprint registers a value change as change.
+     */
+    readonly value?: string;
     /** Href for links; present when the page is same-origin. */
     readonly href?: string;
     readonly children: readonly BrowserNode[];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "dsh-browser-playwright-codex",
-  "version": "0.7.8",
+  "version": "0.8.1",
   "description": "Browser automation for DeepSeek Harness, one plugin: fill in forms, scrape lists and work across tabs in real pages, using the logins you already have. Runs in a visible window that stays quiet when minimised; native dialogs are parked instead of confirmed, CAPTCHA pages are flagged with a note that says stop and hand the page back, password / one-time-code / card fields are refused by browser_fill, and the Settings card (enable/disable, page-reading mode, capability gates) now lives inside this package.",
   "license": "MIT",
   "type": "module",
@@ -44,6 +44,8 @@
     "CHANGELOG.md",
     "SECURITY.md",
     "NOTICE.md",
+    "assets",
+    "!assets/swarm-rebuild.upstream.js",
     "locale",
     "scripts/doctor.mjs",
     "scripts/dsh-browser-verify.mjs"
@@ -57,10 +59,12 @@
     "verify": "npm run build && node scripts/dsh-browser-verify.mjs",
     "verify:live": "npm run build && node scripts/dsh-browser-verify.mjs --live",
     "doctor": "node scripts/doctor.mjs",
+    "swarm:patch": "node scripts/swarm-patch.mjs",
+    "swarm:check": "node scripts/swarm-patch.mjs --check",
     "cost": "node node_modules/tsx/dist/cli.mjs scripts/measure-prompt-cost.mjs",
     "format": "node node_modules/prettier/bin/prettier.cjs --write \"src/**/*.ts\" \"tests/**/*.ts\" \"scripts/**/*.mjs\"",
     "format:check": "node node_modules/prettier/bin/prettier.cjs --check \"src/**/*.ts\" \"tests/**/*.ts\" \"scripts/**/*.mjs\"",
-    "prepublishOnly": "npm run build && npm run typecheck && npm test && npm run verify && npm run preflight",
+    "prepublishOnly": "npm run build && npm run typecheck && npm test && npm run swarm:check && npm run verify && npm run preflight",
     "preflight": "node scripts/preflight.mjs ."
   },
   "dependencies": {
@@ -68,19 +72,20 @@
     "playwright-core": "~1.62.1"
   },
   "peerDependencies": {
-    "@deepseek-ai/cordis": ">=4.0.4",
-    "@deepseek-ai/dsh-llm": ">=0.2.0-rc.1",
-    "@deepseek-ai/dsh-tools": ">=0.2.0-rc.1"
+    "@deepseek-ai/cordis": ">=4.1.0",
+    "@deepseek-ai/dsh-llm": ">=0.2.2-rc.1",
+    "@deepseek-ai/dsh-tools": ">=0.2.2-rc.1"
   },
   "devDependencies": {
-    "@deepseek-ai/cordis": "4.0.4",
-    "@deepseek-ai/dsh-agent": "0.2.0-rc.1",
-    "@deepseek-ai/dsh-attachment": "0.2.0-rc.1",
-    "@deepseek-ai/dsh-llm": "0.2.0-rc.1",
-    "@deepseek-ai/dsh-session": "0.2.0-rc.1",
-    "@deepseek-ai/dsh-system-prompt": "0.2.0-rc.1",
-    "@deepseek-ai/dsh-tools": "0.2.0-rc.1",
-    "@deepseek-ai/schemastery": "3.18.4",
+    "@deepseek-ai/cordis": "4.1.0",
+    "@deepseek-ai/dsh-agent": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-attachment": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-llm": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-session": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-system-prompt": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-tools": "0.2.2-rc.1",
+    "@deepseek-ai/dsh-tool-executor-native": "0.2.2-rc.1",
+    "@deepseek-ai/schemastery": "3.19.0",
     "@types/node": "^24.13.3",
     "prettier": "^3.9.9",
     "tsx": "^4.20.0",