langflower 0.1.0 → 0.1.2

package/vendor/common-nodes/dist/crawl/crawl/node.d.ts

@@ -1,99 +0,0 @@
-/**
- * BFS crawl from a seed URL with depth / page caps.
- *
- * Uses SSRF-guarded `createWebFetch` and optionally `createCrawlContext.savePage`.
- * Default rate/budget control is hard caps (`maxPages`, `maxDepth`) — not a
- * global QPS limiter. CAPTCHA / browser automation are out of scope.
- * Algorithm lives in `@langflower/tools/run-bfs-crawl`.
- */
-export declare const crawlNode: {
-    readonly emitOncePerActivation: boolean;
-    readonly stopsRun: boolean;
-    readonly chatEntry: boolean;
-    readonly uiSchema: readonly [{
-        readonly field: "maxDepth";
-        readonly type: "number";
-        readonly label: "Max depth";
-        readonly default: 1;
-    }, {
-        readonly field: "maxPages";
-        readonly type: "number";
-        readonly label: "Max pages";
-        readonly default: 8;
-    }, {
-        readonly field: "sameHostOnly";
-        readonly type: "boolean";
-        readonly label: "Same host only";
-        readonly default: true;
-    }, {
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout per page (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max bytes per page";
-        readonly default: 5000000;
-    }];
-    readonly bypassPorts: Record<string, import("@langflower/node-sdk").WireType>;
-    readonly inputsConfigs: readonly [import("@langflower/node-sdk").InputPortMeta<import("@langflower/node-sdk").ExecutionContext<readonly [{
-        readonly field: "maxDepth";
-        readonly type: "number";
-        readonly label: "Max depth";
-        readonly default: 1;
-    }, {
-        readonly field: "maxPages";
-        readonly type: "number";
-        readonly label: "Max pages";
-        readonly default: 8;
-    }, {
-        readonly field: "sameHostOnly";
-        readonly type: "boolean";
-        readonly label: "Same host only";
-        readonly default: true;
-    }, {
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout per page (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max bytes per page";
-        readonly default: 5000000;
-    }], Record<string, never>>>, ...import("@langflower/node-sdk").InputPortMeta<unknown>[]];
-    readonly outputsConfigs: (import("@langflower/node-sdk").OutputParams & import("@langflower/node-sdk").PortMeta)[];
-    readonly getInstance: () => import("@langflower/node-sdk").ReactiveNodeInstance<readonly [{
-        readonly field: "maxDepth";
-        readonly type: "number";
-        readonly label: "Max depth";
-        readonly default: 1;
-    }, {
-        readonly field: "maxPages";
-        readonly type: "number";
-        readonly label: "Max pages";
-        readonly default: 8;
-    }, {
-        readonly field: "sameHostOnly";
-        readonly type: "boolean";
-        readonly label: "Same host only";
-        readonly default: true;
-    }, {
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout per page (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max bytes per page";
-        readonly default: 5000000;
-    }], Record<string, never>>;
-    readonly paletteSecondary?: true;
-    readonly description?: string;
-    readonly type: string;
-    readonly displayName: string;
-    readonly category: string | undefined;
-    readonly icon: string | undefined;
-};

package/vendor/common-nodes/dist/crawl/crawl/node.js

@@ -1,128 +0,0 @@
-import { defineReactiveNode, withLoading } from '@langflower/node-sdk';
-import { createCrawlContext } from '@langflower/tools/create-crawl-context';
-import { createWebFetch } from '@langflower/tools/create-web-fetch';
-import { runBfsCrawl } from '@langflower/tools/run-bfs-crawl';
-import { from, mergeMap, throwError } from 'rxjs';
-import { getRunHostServices } from '../../ai/features/run-host-services.js';
-const normalizeLimit = (value, fallback, max) => {
-    const n = Number(value);
-    if (!Number.isFinite(n) || n < 0) {
-        return fallback;
-    }
-    return Math.min(Math.floor(n), max);
-};
-/**
- * BFS crawl from a seed URL with depth / page caps.
- *
- * Uses SSRF-guarded `createWebFetch` and optionally `createCrawlContext.savePage`.
- * Default rate/budget control is hard caps (`maxPages`, `maxDepth`) — not a
- * global QPS limiter. CAPTCHA / browser automation are out of scope.
- * Algorithm lives in `@langflower/tools/run-bfs-crawl`.
- */
-export const crawlNode = defineReactiveNode({
-    type: 'common-crawl',
-    displayName: 'Crawl',
-    category: 'Crawl',
-    paletteSecondary: true,
-    description: `
-Start from a URL and follow links up to a depth and page cap. Pages are saved as they are visited.
-
-Typical uses:
-- Research a site without clicking by hand
-- Collect a small corpus for an LLM
-`.trim(),
-    uiSchema: [
-        {
-            field: 'maxDepth',
-            type: 'number',
-            label: 'Max depth',
-            default: 1,
-        },
-        {
-            field: 'maxPages',
-            type: 'number',
-            label: 'Max pages',
-            default: 8,
-        },
-        {
-            field: 'sameHostOnly',
-            type: 'boolean',
-            label: 'Same host only',
-            default: true,
-        },
-        {
-            field: 'timeoutMs',
-            type: 'number',
-            label: 'Timeout per page (ms)',
-            default: 30_000,
-        },
-        {
-            field: 'maxBytes',
-            type: 'number',
-            label: 'Max bytes per page',
-            default: 5_000_000,
-        },
-    ],
-    bind(ctx, { makeInput, configureOutput, combineInputs }) {
-        const startUrl = makeInput('startUrl', {
-            name: 'startUrl',
-            wireType: 'string',
-            inline: 'text',
-            required: true,
-        });
-        const pages$ = combineInputs([startUrl, ctx], ([rawUrl, ec]) => ({
-            startUrl: String(rawUrl ?? '').trim(),
-            ec,
-        }))
-            .pipe(withLoading())
-            .pipeValue(mergeMap(({ startUrl: seed, ec }) => {
-            if (seed.length === 0) {
-                return throwError(() => new Error('Crawl requires a non-empty startUrl.'));
-            }
-            const hostServices = getRunHostServices(ec);
-            const webFetch = createWebFetch({
-                ...(hostServices?.allowedHosts !== undefined
-                    ? { allowedHosts: hostServices.allowedHosts }
-                    : {}),
-            });
-            const crawl = createCrawlContext(ec.projectDir, ec.runId);
-            return from(runBfsCrawl({
-                startUrl: seed,
-                maxDepth: normalizeLimit(ec.params.maxDepth, 1, 5),
-                maxPages: normalizeLimit(ec.params.maxPages, 8, 50),
-                sameHostOnly: ec.params.sameHostOnly !== false,
-                timeoutMs: normalizeLimit(ec.params.timeoutMs, 30_000, 120_000),
-                maxBytes: normalizeLimit(ec.params.maxBytes, 5_000_000, 20_000_000),
-                failureMode: 'skip',
-                enqueueBudget: 'maxPages',
-                webFetch,
-                savePage: crawl.savePage,
-            }).then((crawled) => crawled.flatMap((page) => {
-                if (!page.ok) {
-                    return [];
-                }
-                return [
-                    {
-                        url: page.url,
-                        text: page.text,
-                        status: page.status,
-                        ...(page.title !== undefined
-                            ? { title: page.title }
-                            : {}),
-                        ...(page.savedPath !== undefined
-                            ? { savedPath: page.savedPath }
-                            : {}),
-                    },
-                ];
-            })));
-        }));
-        return {
-            inputs: [startUrl],
-            outputs: [
-                configureOutput('pages', pages$, {
-                    wireType: 'json',
-                }),
-            ],
-        };
-    },
-});

package/vendor/common-nodes/dist/crawl/crawl-tools/node.d.ts

@@ -1,20 +0,0 @@
-/**
- * Emits Crawl ToolHandles. Invokers come from `@langflower/tools`; no harness
- * tool-id registry participates in agent inventory.
- */
-export declare const crawlToolsNode: {
-    readonly emitOncePerActivation: boolean;
-    readonly stopsRun: boolean;
-    readonly chatEntry: boolean;
-    readonly uiSchema: readonly [];
-    readonly bypassPorts: Record<string, import("@langflower/node-sdk").WireType>;
-    readonly inputsConfigs: readonly [import("@langflower/node-sdk").InputPortMeta<import("@langflower/node-sdk").ExecutionContext<readonly [], Record<string, never>>>, ...import("@langflower/node-sdk").InputPortMeta<unknown>[]];
-    readonly outputsConfigs: (import("@langflower/node-sdk").OutputParams & import("@langflower/node-sdk").PortMeta)[];
-    readonly getInstance: () => import("@langflower/node-sdk").ReactiveNodeInstance<readonly [], Record<string, never>>;
-    readonly paletteSecondary?: true;
-    readonly description?: string;
-    readonly type: string;
-    readonly displayName: string;
-    readonly category: string | undefined;
-    readonly icon: string | undefined;
-};

package/vendor/common-nodes/dist/crawl/crawl-tools/node.js

@@ -1,17 +0,0 @@
-import { defineToolRegistrations } from '@langflower/node-sdk';
-import { CRAWL_TOOL_CONFIGS } from '@langflower/tools/domain-tool-configs';
-/**
- * Emits Crawl ToolHandles. Invokers come from `@langflower/tools`; no harness
- * tool-id registry participates in agent inventory.
- */
-export const crawlToolsNode = defineToolRegistrations({
-    type: 'common-crawl-tools',
-    displayName: 'Crawl Tools',
-    category: 'Tools',
-    description: `
-Give an agent fetch, extract-links, save-page, and crawl tools.
-
-Wire **tools** into an LLM so it can research the web from chat.
-`.trim(),
-    tools: CRAWL_TOOL_CONFIGS,
-});

package/vendor/common-nodes/dist/crawl/extract-links/node.d.ts

@@ -1,22 +0,0 @@
-/**
- * Extract absolute links from an HTML document.
- *
- * **Use when:** expanding a seed page into crawl candidates before Merge or Crawl.
- * Pure / offline — no network I/O.
- */
-export declare const extractLinksNode: {
-    readonly emitOncePerActivation: boolean;
-    readonly stopsRun: boolean;
-    readonly chatEntry: boolean;
-    readonly uiSchema: readonly [];
-    readonly bypassPorts: Record<string, import("@langflower/node-sdk").WireType>;
-    readonly inputsConfigs: readonly [import("@langflower/node-sdk").InputPortMeta<import("@langflower/node-sdk").ExecutionContext<readonly [], Record<string, never>>>, ...import("@langflower/node-sdk").InputPortMeta<unknown>[]];
-    readonly outputsConfigs: (import("@langflower/node-sdk").OutputParams & import("@langflower/node-sdk").PortMeta)[];
-    readonly getInstance: () => import("@langflower/node-sdk").ReactiveNodeInstance<readonly [], Record<string, never>>;
-    readonly paletteSecondary?: true;
-    readonly description?: string;
-    readonly type: string;
-    readonly displayName: string;
-    readonly category: string | undefined;
-    readonly icon: string | undefined;
-};

package/vendor/common-nodes/dist/crawl/extract-links/node.js

@@ -1,45 +0,0 @@
-import { defineReactiveNode } from '@langflower/node-sdk';
-import { extractLinks } from '@langflower/tools/html';
-/**
- * Extract absolute links from an HTML document.
- *
- * **Use when:** expanding a seed page into crawl candidates before Merge or Crawl.
- * Pure / offline — no network I/O.
- */
-export const extractLinksNode = defineReactiveNode({
-    type: 'common-extract-links',
-    displayName: 'Extract Links',
-    category: 'Crawl',
-    paletteSecondary: true,
-    description: `
-Pull absolute URLs from HTML.
-
-Typical uses:
-- List outbound links after Fetch URL
-- Feed those URLs into Crawl or another fetch
-`.trim(),
-    uiSchema: [],
-    bind(_ctx, { makeInput, configureOutput, combineInputs }) {
-        const html = makeInput('html', {
-            name: 'html',
-            wireType: 'string',
-            inline: 'text-multiline',
-            required: true,
-        });
-        const baseUrl = makeInput('baseUrl', {
-            name: 'baseUrl',
-            wireType: 'string',
-            inline: 'text',
-            required: true,
-        });
-        const links$ = combineInputs([html, baseUrl], ([rawHtml, rawBase]) => extractLinks(String(rawHtml ?? ''), String(rawBase ?? '')));
-        return {
-            inputs: [html, baseUrl],
-            outputs: [
-                configureOutput('links', links$, {
-                    wireType: 'json',
-                }),
-            ],
-        };
-    },
-});

package/vendor/common-nodes/dist/crawl/fetch-url/node.d.ts

@@ -1,56 +0,0 @@
-/**
- * HTTP GET a URL via server `ctx.harness.webFetch` (SSRF guards), then extract
- * plain text from HTML.
- *
- * **Use when:** a research branch needs one page’s body without a full BFS crawl.
- *
- * Safety: private/loopback/link-local targets are blocked server-side; optional
- * `harness.allowedHosts` allowlist in `langflower.jsonc`. Prefer offline mocks
- * in tests — never call the public internet from unit tests.
- */
-export declare const fetchUrlNode: {
-    readonly emitOncePerActivation: boolean;
-    readonly stopsRun: boolean;
-    readonly chatEntry: boolean;
-    readonly uiSchema: readonly [{
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max response bytes";
-        readonly default: 5000000;
-    }];
-    readonly bypassPorts: Record<string, import("@langflower/node-sdk").WireType>;
-    readonly inputsConfigs: readonly [import("@langflower/node-sdk").InputPortMeta<import("@langflower/node-sdk").ExecutionContext<readonly [{
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max response bytes";
-        readonly default: 5000000;
-    }], Record<string, never>>>, ...import("@langflower/node-sdk").InputPortMeta<unknown>[]];
-    readonly outputsConfigs: (import("@langflower/node-sdk").OutputParams & import("@langflower/node-sdk").PortMeta)[];
-    readonly getInstance: () => import("@langflower/node-sdk").ReactiveNodeInstance<readonly [{
-        readonly field: "timeoutMs";
-        readonly type: "number";
-        readonly label: "Timeout (ms)";
-        readonly default: 30000;
-    }, {
-        readonly field: "maxBytes";
-        readonly type: "number";
-        readonly label: "Max response bytes";
-        readonly default: 5000000;
-    }], Record<string, never>>;
-    readonly paletteSecondary?: true;
-    readonly description?: string;
-    readonly type: string;
-    readonly displayName: string;
-    readonly category: string | undefined;
-    readonly icon: string | undefined;
-};

package/vendor/common-nodes/dist/crawl/fetch-url/node.js

@@ -1,89 +0,0 @@
-import { defineReactiveNode, withLoading } from '@langflower/node-sdk';
-import { createWebFetch } from '@langflower/tools/create-web-fetch';
-import { from, map, mergeMap, throwError } from 'rxjs';
-import { htmlToText } from '@langflower/tools/html';
-import { getRunHostServices } from '../../ai/features/run-host-services.js';
-/**
- * HTTP GET a URL via server `ctx.harness.webFetch` (SSRF guards), then extract
- * plain text from HTML.
- *
- * **Use when:** a research branch needs one page’s body without a full BFS crawl.
- *
- * Safety: private/loopback/link-local targets are blocked server-side; optional
- * `harness.allowedHosts` allowlist in `langflower.jsonc`. Prefer offline mocks
- * in tests — never call the public internet from unit tests.
- */
-export const fetchUrlNode = defineReactiveNode({
-    type: 'common-fetch-url',
-    displayName: 'Fetch URL',
-    category: 'Crawl',
-    paletteSecondary: true,
-    description: `
-Download a page and get HTML plus readable text.
-
-Typical uses:
-- Pull an article before Extract Links or an LLM
-- Seed a Crawl
-`.trim(),
-    uiSchema: [
-        {
-            field: 'timeoutMs',
-            type: 'number',
-            label: 'Timeout (ms)',
-            default: 30_000,
-        },
-        {
-            field: 'maxBytes',
-            type: 'number',
-            label: 'Max response bytes',
-            default: 5_000_000,
-        },
-    ],
-    bind(ctx, { makeInput, configureOutput, combineInputs }) {
-        const url = makeInput('url', {
-            name: 'url',
-            wireType: 'string',
-            inline: 'text',
-            required: true,
-        });
-        const fetched$ = combineInputs([url, ctx], ([rawUrl, ec]) => ({
-            rawUrl: String(rawUrl ?? '').trim(),
-            ec,
-        }))
-            .pipe(withLoading())
-            .pipeValue(mergeMap(({ rawUrl, ec }) => {
-            if (rawUrl.length === 0) {
-                return throwError(() => new Error('Fetch URL requires a non-empty url.'));
-            }
-            const allowedHosts = getRunHostServices(ec)?.allowedHosts;
-            const webFetch = createWebFetch({
-                ...(allowedHosts !== undefined ? { allowedHosts } : {}),
-            });
-            const timeoutMs = Number(ec.params.timeoutMs ?? 30_000);
-            const maxBytes = Number(ec.params.maxBytes ?? 5_000_000);
-            return from(webFetch({
-                url: rawUrl,
-                timeoutMs,
-                maxBytes,
-            })).pipe(map((result) => {
-                if (!result.ok && result.body.length === 0) {
-                    throw new Error(result.error ??
-                        `Fetch failed for ${rawUrl}`);
-                }
-                return {
-                    html: result.body,
-                    text: htmlToText(result.body),
-                    status: result.status,
-                };
-            }));
-        }));
-        return {
-            inputs: [url],
-            outputs: [
-                configureOutput('text', fetched$.pipeValue(map((page) => page.text)), { wireType: 'string' }),
-                configureOutput('html', fetched$.pipeValue(map((page) => page.html)), { wireType: 'string' }),
-                configureOutput('status', fetched$.pipeValue(map((page) => page.status)), { wireType: 'number' }),
-            ],
-        };
-    },
-});

package/vendor/common-nodes/dist/crawl/html/extract-links.d.ts

@@ -1,3 +0,0 @@
-export declare function extractLinks(html: string, baseUrl: string): readonly string[];
-export declare function isSameHost(leftUrl: string, rightUrl: string): boolean;
-//# sourceMappingURL=extract-links.d.ts.map

package/vendor/common-nodes/dist/crawl/html/extract-links.js

@@ -1,36 +0,0 @@
-function resolveLink(baseUrl, href) {
-    const trimmed = href.trim();
-    if (trimmed.length === 0 ||
-        trimmed.startsWith('#') ||
-        trimmed.startsWith('mailto:') ||
-        trimmed.startsWith('javascript:')) {
-        return null;
-    }
-    try {
-        return new URL(trimmed, baseUrl).toString();
-    }
-    catch {
-        return null;
-    }
-}
-export function extractLinks(html, baseUrl) {
-    const links = new Set();
-    const pattern = /\shref\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s>]+))/gi;
-    for (const match of html.matchAll(pattern)) {
-        const href = match[1] ?? match[2] ?? match[3] ?? '';
-        const resolved = resolveLink(baseUrl, href);
-        if (resolved !== null) {
-            links.add(resolved);
-        }
-    }
-    return [...links].sort((left, right) => left.localeCompare(right));
-}
-export function isSameHost(leftUrl, rightUrl) {
-    try {
-        return new URL(leftUrl).hostname === new URL(rightUrl).hostname;
-    }
-    catch {
-        return false;
-    }
-}
-//# sourceMappingURL=extract-links.js.map

package/vendor/common-nodes/dist/crawl/html/html-to-text.d.ts

@@ -1,3 +0,0 @@
-export declare function htmlToText(html: string): string;
-export declare function extractHtmlTitle(html: string): string | undefined;
-//# sourceMappingURL=html-to-text.d.ts.map

package/vendor/common-nodes/dist/crawl/html/html-to-text.js

@@ -1,36 +0,0 @@
-const BLOCK_TAGS = /<(script|style|noscript|svg|iframe)[^>]*>[\s\S]*?<\/\1>/gi;
-const TAG_PATTERN = /<[^>]+>/g;
-const WHITESPACE_PATTERN = /\s+/g;
-const ENTITY_MAP = {
-    amp: '&',
-    lt: '<',
-    gt: '>',
-    quot: '"',
-    apos: "'",
-    nbsp: ' ',
-};
-function decodeBasicEntities(text) {
-    return text.replace(/&(#x?[0-9a-fA-F]+|\w+);/g, (match, entity) => {
-        if (entity.startsWith('#x')) {
-            const code = Number.parseInt(entity.slice(2), 16);
-            return Number.isFinite(code) ? String.fromCodePoint(code) : match;
-        }
-        if (entity.startsWith('#')) {
-            const code = Number.parseInt(entity.slice(1), 10);
-            return Number.isFinite(code) ? String.fromCodePoint(code) : match;
-        }
-        return ENTITY_MAP[entity] ?? match;
-    });
-}
-export function htmlToText(html) {
-    const withoutBlocks = html.replace(BLOCK_TAGS, ' ');
-    const withoutTags = withoutBlocks.replace(TAG_PATTERN, ' ');
-    const decoded = decodeBasicEntities(withoutTags);
-    return decoded.replace(WHITESPACE_PATTERN, ' ').trim();
-}
-export function extractHtmlTitle(html) {
-    const match = /<title[^>]*>([\s\S]*?)<\/title>/i.exec(html);
-    const title = match?.[1]?.replace(TAG_PATTERN, ' ').trim();
-    return title && title.length > 0 ? decodeBasicEntities(title) : undefined;
-}
-//# sourceMappingURL=html-to-text.js.map

package/vendor/common-nodes/dist/crawl/html/index.d.ts

@@ -1,3 +0,0 @@
-export { extractLinks, isSameHost } from './extract-links.js';
-export { extractHtmlTitle, htmlToText } from './html-to-text.js';
-//# sourceMappingURL=index.d.ts.map

package/vendor/common-nodes/dist/crawl/html/index.js

@@ -1,3 +0,0 @@
-export { extractLinks, isSameHost } from './extract-links.js';
-export { extractHtmlTitle, htmlToText } from './html-to-text.js';
-//# sourceMappingURL=index.js.map

package/vendor/common-nodes/dist/crawl/save-page/node.d.ts

@@ -1,21 +0,0 @@
-/**
- * Persist a fetched page under `.langflower/crawl/{runId}/` via `createCrawlContext`.
- *
- * **Use when:** you want durable research artifacts for later Merge / Preview.
- */
-export declare const savePageNode: {
-    readonly emitOncePerActivation: boolean;
-    readonly stopsRun: boolean;
-    readonly chatEntry: boolean;
-    readonly uiSchema: readonly [];
-    readonly bypassPorts: Record<string, import("@langflower/node-sdk").WireType>;
-    readonly inputsConfigs: readonly [import("@langflower/node-sdk").InputPortMeta<import("@langflower/node-sdk").ExecutionContext<readonly [], Record<string, never>>>, ...import("@langflower/node-sdk").InputPortMeta<unknown>[]];
-    readonly outputsConfigs: (import("@langflower/node-sdk").OutputParams & import("@langflower/node-sdk").PortMeta)[];
-    readonly getInstance: () => import("@langflower/node-sdk").ReactiveNodeInstance<readonly [], Record<string, never>>;
-    readonly paletteSecondary?: true;
-    readonly description?: string;
-    readonly type: string;
-    readonly displayName: string;
-    readonly category: string | undefined;
-    readonly icon: string | undefined;
-};

package/vendor/common-nodes/dist/crawl/save-page/node.js

@@ -1,71 +0,0 @@
-import { defineReactiveNode, withLoading } from '@langflower/node-sdk';
-import { createCrawlContext } from '@langflower/tools/create-crawl-context';
-import { from, mergeMap, throwError } from 'rxjs';
-import { extractHtmlTitle } from '@langflower/tools/html';
-/**
- * Persist a fetched page under `.langflower/crawl/{runId}/` via `createCrawlContext`.
- *
- * **Use when:** you want durable research artifacts for later Merge / Preview.
- */
-export const savePageNode = defineReactiveNode({
-    type: 'common-save-page',
-    displayName: 'Save Page',
-    category: 'Crawl',
-    paletteSecondary: true,
-    description: `
-Store a fetched page in this run's crawl folder and emit where it was saved.
-
-Typical uses:
-- Keep HTML/text for later review
-- Persist a page before the next crawl hop
-`.trim(),
-    uiSchema: [],
-    bind(ctx, { makeInput, configureOutput, combineInputs }) {
-        const url = makeInput('url', {
-            name: 'url',
-            wireType: 'string',
-            inline: 'text',
-            required: true,
-        });
-        const html = makeInput('html', {
-            name: 'html',
-            wireType: 'string',
-            inline: 'text-multiline',
-            defaultValue: '',
-        });
-        const text = makeInput('text', {
-            name: 'text',
-            wireType: 'string',
-            inline: 'text-multiline',
-            defaultValue: '',
-        });
-        const saved$ = combineInputs([url, html, text, ctx], ([rawUrl, rawHtml, rawText, ec]) => ({
-            url: String(rawUrl ?? '').trim(),
-            html: String(rawHtml ?? ''),
-            text: String(rawText ?? ''),
-            ec,
-        }))
-            .pipe(withLoading())
-            .pipeValue(mergeMap(({ url: pageUrl, html: pageHtml, text: pageText, ec }) => {
-            if (pageUrl.length === 0) {
-                return throwError(() => new Error('Save Page requires a non-empty url.'));
-            }
-            const crawl = createCrawlContext(ec.projectDir, ec.runId);
-            const title = extractHtmlTitle(pageHtml);
-            return from(crawl.savePage({
-                url: pageUrl,
-                html: pageHtml,
-                text: pageText,
-                ...(title !== undefined ? { title } : {}),
-            }));
-        }));
-        return {
-            inputs: [url, html, text],
-            outputs: [
-                configureOutput('saved', saved$, {
-                    wireType: 'json',
-                }),
-            ],
-        };
-    },
-});