@staticn0va/wigolo 0.1.0 → 0.1.2

package/dist/server/tool-schemas.js

@@ -0,0 +1,464 @@
+const FETCH_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    url: { type: "string", description: "URL to fetch" },
+    render_js: {
+      type: "string",
+      enum: ["auto", "always", "never"],
+      description: "JavaScript rendering mode (default: auto)"
+    },
+    use_auth: {
+      type: "boolean",
+      description: "Use stored auth credentials (default: false)"
+    },
+    max_chars: {
+      type: "number",
+      description: "Maximum characters to return (hard slice)"
+    },
+    max_content_chars: {
+      type: "number",
+      description: "Smart truncate markdown to N chars at paragraph/heading boundary with [... content truncated] marker. Preferred over max_chars for AI agents."
+    },
+    section: {
+      type: "string",
+      description: "Extract a specific section by heading text"
+    },
+    section_index: {
+      type: "number",
+      description: "Index of the section match (default: 0)"
+    },
+    screenshot: {
+      type: "boolean",
+      description: "Capture a screenshot (default: false)"
+    },
+    headers: {
+      type: "object",
+      description: "Additional HTTP headers",
+      additionalProperties: { type: "string" }
+    },
+    force_refresh: {
+      type: "boolean",
+      description: "Bypass cache and fetch fresh content from the network. Use for rapidly changing pages (news, changelogs, dashboards)."
+    },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    },
+    actions: {
+      type: "array",
+      description: "Sequential browser actions to perform before extracting content. When present, forces browser rendering (bypasses HTTP-first routing).",
+      items: {
+        type: "object",
+        properties: {
+          type: {
+            type: "string",
+            enum: ["click", "type", "wait", "wait_for", "scroll", "screenshot"],
+            description: "Action type"
+          },
+          selector: {
+            type: "string",
+            description: "CSS selector (required for click, type, wait_for)"
+          },
+          text: {
+            type: "string",
+            description: "Text to type (required for type action)"
+          },
+          ms: {
+            type: "number",
+            description: "Milliseconds to wait (required for wait action)"
+          },
+          timeout: {
+            type: "number",
+            description: "Timeout in ms for wait_for action (default: 5000)"
+          },
+          direction: {
+            type: "string",
+            enum: ["down", "up"],
+            description: "Scroll direction (required for scroll action)"
+          },
+          amount: {
+            type: "number",
+            description: "Scroll amount in pixels (default: viewport height)"
+          }
+        },
+        required: ["type"]
+      }
+    },
+    mode: {
+      type: "string",
+      enum: ["cache", "default", "stealth"],
+      description: "cache=HTTP-only, accepts stale cache. default=standard fetch with JS detection. stealth=full browser render."
+    }
+  },
+  required: ["url"]
+};
+const SEARCH_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    query: {
+      oneOf: [
+        { type: "string", description: "Search query" },
+        {
+          type: "array",
+          items: { type: "string" },
+          description: "Array of query variants to search in parallel, deduplicate, and rerank"
+        }
+      ],
+      description: "Search query \u2014 a single string or array of query variants for parallel multi-query search"
+    },
+    max_results: { type: "number", description: "Max results to return (default 5, max 20)" },
+    max_fetches: { type: "number", description: "Cap on how many top-ranked results have their page content fetched. Defaults to max_results. Set lower (e.g. 3) to keep snippet-only listings cheap and only deep-read the most relevant." },
+    include_content: { type: "boolean", description: "Fetch full content for results (default true)" },
+    content_max_chars: { type: "number", description: "Max chars per result content at extraction (default 30000)" },
+    max_content_chars: { type: "number", description: "Smart-truncate each result markdown at paragraph boundary with marker (e.g. 3000 for compact context)" },
+    max_total_chars: { type: "number", description: "Max total chars across all results (default 50000)" },
+    time_range: { type: "string", enum: ["day", "week", "month", "year"], description: "Time range filter" },
+    search_engines: { type: "array", items: { type: "string" }, description: "Override engine selection" },
+    language: { type: "string", description: "Language preference" },
+    include_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: 'Only return results from these domains (e.g. ["react.dev", "github.com"])'
+    },
+    exclude_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: "Never return results from these domains"
+    },
+    from_date: {
+      type: "string",
+      description: "ISO date (YYYY-MM-DD) \u2014 only return results published after this date"
+    },
+    to_date: {
+      type: "string",
+      description: "ISO date (YYYY-MM-DD) \u2014 only return results published before this date"
+    },
+    category: {
+      type: "string",
+      enum: ["general", "news", "code", "docs", "papers", "images"],
+      description: "Category of search (general, news, code, docs, papers, images)"
+    },
+    format: {
+      type: "string",
+      enum: ["answer", "stream_answer"],
+      description: "LLM-synthesis modes only. Omit for default evidence shape. 'answer'/'stream_answer' request sampling synthesis (falls back to evidence). Retired values 'full'/'context'/'highlights' reject with a migration error."
+    },
+    max_highlights: {
+      type: "number",
+      description: "Maximum highlights to return (default 10). Highlights are 1-3 sentence passages scored by relevance to the query."
+    },
+    force_refresh: {
+      type: "boolean",
+      description: "Bypass all caches (search results and page content). Use when you need the most current information."
+    },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    },
+    mode: {
+      type: "string",
+      enum: ["cache", "default", "stealth"],
+      description: "cache=single-engine, no rerank, stale cache ok. default=standard multi-engine search. stealth=full browser for JS-heavy result pages."
+    },
+    agent_context: {
+      type: "object",
+      description: "Optional agent context for ranking + dedup. text is concatenated with the query before embedding; recent_urls are dropped from results.",
+      properties: {
+        text: { type: "string", description: "Surrounding code / prior turn / task framing." },
+        recent_urls: { type: "array", items: { type: "string" }, description: "URLs the agent has already seen." },
+        intent: { type: "string", description: "One-line task framing. Used when text is omitted." }
+      },
+      additionalProperties: false
+    }
+  },
+  required: ["query"]
+};
+const CRAWL_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    url: { type: "string", description: "Seed URL to start crawling from" },
+    max_depth: { type: "number", description: "Maximum link depth from seed (default: 2)" },
+    max_pages: { type: "number", description: "Maximum pages to crawl (default: 20)" },
+    strategy: {
+      type: "string",
+      enum: ["bfs", "dfs", "sitemap", "map"],
+      description: "Crawl strategy: bfs (breadth-first), dfs (depth-first), sitemap (use sitemap.xml), map (URL-only discovery \u2014 returns list of URLs without content, faster than full crawl)"
+    },
+    include_patterns: {
+      type: "array",
+      items: { type: "string" },
+      description: "URL regex whitelist \u2014 only crawl matching URLs"
+    },
+    exclude_patterns: {
+      type: "array",
+      items: { type: "string" },
+      description: "URL regex blacklist \u2014 skip matching URLs"
+    },
+    use_auth: { type: "boolean", description: "Use stored auth credentials (default: false)" },
+    extract_links: { type: "boolean", description: "Return link graph between pages (default: false)" },
+    max_total_chars: { type: "number", description: "Max total chars across all pages (default: 100000)" },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    }
+  },
+  required: ["url"]
+};
+const CACHE_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    query: { type: "string", description: "Full-text search over cached content" },
+    url_pattern: {
+      type: "string",
+      description: 'Filter by URL glob pattern (e.g., "*example.com*")'
+    },
+    since: {
+      type: "string",
+      description: "ISO date \u2014 only results cached after this date"
+    },
+    clear: {
+      type: "boolean",
+      description: "Clear matching cache entries (requires at least one filter: query, url_pattern, or since)"
+    },
+    stats: {
+      type: "boolean",
+      description: "Return cache statistics (total URLs, size, date range)"
+    },
+    check_changes: {
+      type: "boolean",
+      description: "Re-fetch all matching cached URLs and report which ones have changed. Returns a list of URLs with changed/unchanged status and diff summaries. Use with query or url_pattern to scope which cached entries to check."
+    },
+    mode: {
+      type: "string",
+      enum: ["fts", "hybrid"],
+      description: 'Search strategy when query is provided. "fts" (default) runs keyword-only BM25 over the FTS5 index. "hybrid" additionally runs semantic vector search and fuses both rankings with reciprocal rank fusion for higher-recall lookups; falls back to FTS when the embedding index is empty or unavailable.'
+    },
+    limit: {
+      type: "number",
+      description: "Maximum number of results to return (default 20)."
+    }
+  }
+};
+const EXTRACT_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    url: { type: "string", description: "URL to fetch and extract from" },
+    html: { type: "string", description: "Raw HTML to extract from (url takes priority if both provided)" },
+    mode: {
+      type: "string",
+      enum: ["selector", "tables", "metadata", "schema", "structured"],
+      description: "Extraction mode: selector (CSS), tables (HTML tables), metadata (meta tags + JSON-LD), schema (fields matching a JSON Schema), structured (tables + definition lists + JSON-LD + chart hints + key/value pairs \u2014 one-shot structured brief)"
+    },
+    css_selector: {
+      type: "string",
+      description: 'CSS selector to match (required when mode="selector")'
+    },
+    multiple: {
+      type: "boolean",
+      description: 'Return array of all matches instead of first (default: false, only for mode="selector")'
+    },
+    schema: {
+      type: "object",
+      description: 'JSON Schema defining fields to extract. Field names are matched against page content via CSS classes, ARIA labels, microdata, and JSON-LD. Required when mode="schema".'
+    },
+    named_schema: {
+      type: "string",
+      enum: ["Article", "Recipe", "Product", "CodeSnippet", "Paper", "EventListing"],
+      description: "Extract page data into a strict named schema (heuristic only; no LLM required). Mutually exclusive with `schema`."
+    }
+  }
+};
+const FIND_SIMILAR_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    url: {
+      type: "string",
+      description: "Find pages similar to this URL. The page is fetched (or read from cache) and its content analyzed for key terms."
+    },
+    concept: {
+      type: "string",
+      description: "Find pages related to this concept or topic description. Use when you don't have a specific URL."
+    },
+    max_results: {
+      type: "number",
+      description: "Maximum results to return (default 10, max 50)"
+    },
+    include_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: "Only return results from these domains"
+    },
+    exclude_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: "Never return results from these domains"
+    },
+    include_cache: {
+      type: "boolean",
+      description: "Search local cache for similar pages (default: true)"
+    },
+    include_web: {
+      type: "boolean",
+      description: "Supplement with web search if needed (default: true)"
+    },
+    mode: {
+      type: "string",
+      enum: ["auto", "cache", "web-expansion", "crawl-rank"],
+      default: "auto",
+      description: "Retrieval strategy: cache (local hybrid), web-expansion (key terms + web search), crawl-rank (1-hop crawl from seed URL + embed + cosine rank), or auto."
+    },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    }
+  }
+};
+const RESEARCH_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    question: { type: "string", description: "The research question to investigate" },
+    depth: {
+      type: "string",
+      enum: ["quick", "standard", "comprehensive"],
+      description: "Research depth: quick (~15s), standard (~40s, default), comprehensive (~80s)"
+    },
+    max_sources: {
+      type: "number",
+      description: "Override the default source count for the chosen depth (max 50)"
+    },
+    include_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: "Only search results from these domains"
+    },
+    exclude_domains: {
+      type: "array",
+      items: { type: "string" },
+      description: "Exclude results from these domains"
+    },
+    schema: {
+      type: "object",
+      description: "Optional JSON Schema -- structure the report to extract these fields"
+    },
+    stream: {
+      type: "boolean",
+      description: "Send progress notifications as each research phase completes"
+    },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    }
+  },
+  required: ["question"]
+};
+const AGENT_TOOL_SCHEMA = {
+  type: "object",
+  properties: {
+    prompt: {
+      type: "string",
+      description: "Natural-language description of what data to gather"
+    },
+    urls: {
+      type: "array",
+      items: { type: "string" },
+      description: "Specific URLs to include in the data gathering"
+    },
+    schema: {
+      type: "object",
+      description: "Optional JSON Schema -- extract structured data matching this schema from each page"
+    },
+    max_pages: {
+      type: "number",
+      description: "Maximum pages to fetch (default 10, max 100)"
+    },
+    max_time_ms: {
+      type: "number",
+      description: "Maximum execution time in milliseconds (default 60000)"
+    },
+    stream: {
+      type: "boolean",
+      description: "Send progress notifications as each step completes"
+    },
+    max_tokens_out: {
+      type: "number",
+      description: "Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins."
+    },
+    include_full_markdown: {
+      type: "boolean",
+      description: "Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore."
+    },
+    citation_format: {
+      type: "string",
+      enum: ["numbered", "anthropic_tags", "json"],
+      description: "Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags."
+    }
+  },
+  required: ["prompt"]
+};
+const TOOL_SCHEMAS = {
+  fetch: FETCH_TOOL_SCHEMA,
+  search: SEARCH_TOOL_SCHEMA,
+  crawl: CRAWL_TOOL_SCHEMA,
+  cache: CACHE_TOOL_SCHEMA,
+  extract: EXTRACT_TOOL_SCHEMA,
+  find_similar: FIND_SIMILAR_TOOL_SCHEMA,
+  research: RESEARCH_TOOL_SCHEMA,
+  agent: AGENT_TOOL_SCHEMA
+};
+export {
+  AGENT_TOOL_SCHEMA,
+  CACHE_TOOL_SCHEMA,
+  CRAWL_TOOL_SCHEMA,
+  EXTRACT_TOOL_SCHEMA,
+  FETCH_TOOL_SCHEMA,
+  FIND_SIMILAR_TOOL_SCHEMA,
+  RESEARCH_TOOL_SCHEMA,
+  SEARCH_TOOL_SCHEMA,
+  TOOL_SCHEMAS
+};
+//# sourceMappingURL=tool-schemas.js.map

package/dist/server/tool-schemas.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/server/tool-schemas.ts"],"sourcesContent":["import type { ToolName } from '../instructions.js';\n\nexport type ToolSchema = {\n  type: 'object';\n  properties: Record<string, unknown>;\n  required?: string[];\n};\n\nexport const FETCH_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    url: { type: 'string', description: 'URL to fetch' },\n    render_js: {\n      type: 'string',\n      enum: ['auto', 'always', 'never'],\n      description: 'JavaScript rendering mode (default: auto)',\n    },\n    use_auth: {\n      type: 'boolean',\n      description: 'Use stored auth credentials (default: false)',\n    },\n    max_chars: {\n      type: 'number',\n      description: 'Maximum characters to return (hard slice)',\n    },\n    max_content_chars: {\n      type: 'number',\n      description: 'Smart truncate markdown to N chars at paragraph/heading boundary with [... content truncated] marker. Preferred over max_chars for AI agents.',\n    },\n    section: {\n      type: 'string',\n      description: 'Extract a specific section by heading text',\n    },\n    section_index: {\n      type: 'number',\n      description: 'Index of the section match (default: 0)',\n    },\n    screenshot: {\n      type: 'boolean',\n      description: 'Capture a screenshot (default: false)',\n    },\n    headers: {\n      type: 'object',\n      description: 'Additional HTTP headers',\n      additionalProperties: { type: 'string' },\n    },\n    force_refresh: {\n      type: 'boolean',\n      description: 'Bypass cache and fetch fresh content from the network. Use for rapidly changing pages (news, changelogs, dashboards).',\n    },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n    actions: {\n      type: 'array',\n      description:\n        'Sequential browser actions to perform before extracting content. ' +\n        'When present, forces browser rendering (bypasses HTTP-first routing).',\n      items: {\n        type: 'object',\n        properties: {\n          type: {\n            type: 'string',\n            enum: ['click', 'type', 'wait', 'wait_for', 'scroll', 'screenshot'],\n            description: 'Action type',\n          },\n          selector: {\n            type: 'string',\n            description: 'CSS selector (required for click, type, wait_for)',\n          },\n          text: {\n            type: 'string',\n            description: 'Text to type (required for type action)',\n          },\n          ms: {\n            type: 'number',\n            description: 'Milliseconds to wait (required for wait action)',\n          },\n          timeout: {\n            type: 'number',\n            description: 'Timeout in ms for wait_for action (default: 5000)',\n          },\n          direction: {\n            type: 'string',\n            enum: ['down', 'up'],\n            description: 'Scroll direction (required for scroll action)',\n          },\n          amount: {\n            type: 'number',\n            description: 'Scroll amount in pixels (default: viewport height)',\n          },\n        },\n        required: ['type'],\n      },\n    },\n    mode: {\n      type: 'string',\n      enum: ['cache', 'default', 'stealth'],\n      description: \"cache=HTTP-only, accepts stale cache. default=standard fetch with JS detection. stealth=full browser render.\",\n    },\n  },\n  required: ['url'],\n};\n\nexport const SEARCH_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    query: {\n      oneOf: [\n        { type: 'string', description: 'Search query' },\n        {\n          type: 'array',\n          items: { type: 'string' },\n          description: 'Array of query variants to search in parallel, deduplicate, and rerank',\n        },\n      ],\n      description: 'Search query — a single string or array of query variants for parallel multi-query search',\n    },\n    max_results: { type: 'number', description: 'Max results to return (default 5, max 20)' },\n    max_fetches: { type: 'number', description: 'Cap on how many top-ranked results have their page content fetched. Defaults to max_results. Set lower (e.g. 3) to keep snippet-only listings cheap and only deep-read the most relevant.' },\n    include_content: { type: 'boolean', description: 'Fetch full content for results (default true)' },\n    content_max_chars: { type: 'number', description: 'Max chars per result content at extraction (default 30000)' },\n    max_content_chars: { type: 'number', description: 'Smart-truncate each result markdown at paragraph boundary with marker (e.g. 3000 for compact context)' },\n    max_total_chars: { type: 'number', description: 'Max total chars across all results (default 50000)' },\n    time_range: { type: 'string', enum: ['day', 'week', 'month', 'year'], description: 'Time range filter' },\n    search_engines: { type: 'array', items: { type: 'string' }, description: 'Override engine selection' },\n    language: { type: 'string', description: 'Language preference' },\n    include_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Only return results from these domains (e.g. [\"react.dev\", \"github.com\"])',\n    },\n    exclude_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Never return results from these domains',\n    },\n    from_date: {\n      type: 'string',\n      description: 'ISO date (YYYY-MM-DD) — only return results published after this date',\n    },\n    to_date: {\n      type: 'string',\n      description: 'ISO date (YYYY-MM-DD) — only return results published before this date',\n    },\n    category: {\n      type: 'string',\n      enum: ['general', 'news', 'code', 'docs', 'papers', 'images'],\n      description: 'Category of search (general, news, code, docs, papers, images)',\n    },\n    format: {\n      type: 'string',\n      enum: ['answer', 'stream_answer'],\n      description:\n        \"LLM-synthesis modes only. Omit for default evidence shape. 'answer'/'stream_answer' request sampling synthesis (falls back to evidence). Retired values 'full'/'context'/'highlights' reject with a migration error.\",\n    },\n    max_highlights: {\n      type: 'number',\n      description: \"Maximum highlights to return (default 10). Highlights are 1-3 sentence passages scored by relevance to the query.\",\n    },\n    force_refresh: {\n      type: 'boolean',\n      description: 'Bypass all caches (search results and page content). Use when you need the most current information.',\n    },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n    mode: {\n      type: 'string',\n      enum: ['cache', 'default', 'stealth'],\n      description: \"cache=single-engine, no rerank, stale cache ok. default=standard multi-engine search. stealth=full browser for JS-heavy result pages.\",\n    },\n    agent_context: {\n      type: 'object',\n      description:\n        'Optional agent context for ranking + dedup. text is concatenated with the query before embedding; recent_urls are dropped from results.',\n      properties: {\n        text: { type: 'string', description: 'Surrounding code / prior turn / task framing.' },\n        recent_urls: { type: 'array', items: { type: 'string' }, description: 'URLs the agent has already seen.' },\n        intent: { type: 'string', description: 'One-line task framing. Used when text is omitted.' },\n      },\n      additionalProperties: false,\n    },\n  },\n  required: ['query'],\n};\n\nexport const CRAWL_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    url: { type: 'string', description: 'Seed URL to start crawling from' },\n    max_depth: { type: 'number', description: 'Maximum link depth from seed (default: 2)' },\n    max_pages: { type: 'number', description: 'Maximum pages to crawl (default: 20)' },\n    strategy: {\n      type: 'string',\n      enum: ['bfs', 'dfs', 'sitemap', 'map'],\n      description: 'Crawl strategy: bfs (breadth-first), dfs (depth-first), sitemap (use sitemap.xml), map (URL-only discovery — returns list of URLs without content, faster than full crawl)',\n    },\n    include_patterns: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'URL regex whitelist — only crawl matching URLs',\n    },\n    exclude_patterns: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'URL regex blacklist — skip matching URLs',\n    },\n    use_auth: { type: 'boolean', description: 'Use stored auth credentials (default: false)' },\n    extract_links: { type: 'boolean', description: 'Return link graph between pages (default: false)' },\n    max_total_chars: { type: 'number', description: 'Max total chars across all pages (default: 100000)' },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n  },\n  required: ['url'],\n};\n\nexport const CACHE_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    query: { type: 'string', description: 'Full-text search over cached content' },\n    url_pattern: {\n      type: 'string',\n      description: 'Filter by URL glob pattern (e.g., \"*example.com*\")',\n    },\n    since: {\n      type: 'string',\n      description: 'ISO date — only results cached after this date',\n    },\n    clear: {\n      type: 'boolean',\n      description: 'Clear matching cache entries (requires at least one filter: query, url_pattern, or since)',\n    },\n    stats: {\n      type: 'boolean',\n      description: 'Return cache statistics (total URLs, size, date range)',\n    },\n    check_changes: {\n      type: 'boolean',\n      description:\n        'Re-fetch all matching cached URLs and report which ones have changed. ' +\n        'Returns a list of URLs with changed/unchanged status and diff summaries. ' +\n        'Use with query or url_pattern to scope which cached entries to check.',\n    },\n    mode: {\n      type: 'string',\n      enum: ['fts', 'hybrid'],\n      description:\n        'Search strategy when query is provided. \"fts\" (default) runs keyword-only BM25 over the FTS5 index. ' +\n        '\"hybrid\" additionally runs semantic vector search and fuses both rankings with reciprocal rank fusion ' +\n        'for higher-recall lookups; falls back to FTS when the embedding index is empty or unavailable.',\n    },\n    limit: {\n      type: 'number',\n      description: 'Maximum number of results to return (default 20).',\n    },\n  },\n};\n\nexport const EXTRACT_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    url: { type: 'string', description: 'URL to fetch and extract from' },\n    html: { type: 'string', description: 'Raw HTML to extract from (url takes priority if both provided)' },\n    mode: {\n      type: 'string',\n      enum: ['selector', 'tables', 'metadata', 'schema', 'structured'],\n      description: 'Extraction mode: selector (CSS), tables (HTML tables), metadata (meta tags + JSON-LD), schema (fields matching a JSON Schema), structured (tables + definition lists + JSON-LD + chart hints + key/value pairs — one-shot structured brief)',\n    },\n    css_selector: {\n      type: 'string',\n      description: 'CSS selector to match (required when mode=\"selector\")',\n    },\n    multiple: {\n      type: 'boolean',\n      description: 'Return array of all matches instead of first (default: false, only for mode=\"selector\")',\n    },\n    schema: {\n      type: 'object',\n      description: 'JSON Schema defining fields to extract. Field names are matched against page content via CSS classes, ARIA labels, microdata, and JSON-LD. Required when mode=\"schema\".',\n    },\n    named_schema: {\n      type: 'string',\n      enum: ['Article', 'Recipe', 'Product', 'CodeSnippet', 'Paper', 'EventListing'],\n      description: 'Extract page data into a strict named schema (heuristic only; no LLM required). Mutually exclusive with `schema`.',\n    },\n  },\n};\n\nexport const FIND_SIMILAR_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    url: {\n      type: 'string',\n      description: 'Find pages similar to this URL. The page is fetched (or read from cache) and its content analyzed for key terms.',\n    },\n    concept: {\n      type: 'string',\n      description: 'Find pages related to this concept or topic description. Use when you don\\'t have a specific URL.',\n    },\n    max_results: {\n      type: 'number',\n      description: 'Maximum results to return (default 10, max 50)',\n    },\n    include_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Only return results from these domains',\n    },\n    exclude_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Never return results from these domains',\n    },\n    include_cache: {\n      type: 'boolean',\n      description: 'Search local cache for similar pages (default: true)',\n    },\n    include_web: {\n      type: 'boolean',\n      description: 'Supplement with web search if needed (default: true)',\n    },\n    mode: {\n      type: 'string',\n      enum: ['auto', 'cache', 'web-expansion', 'crawl-rank'],\n      default: 'auto',\n      description: \"Retrieval strategy: cache (local hybrid), web-expansion (key terms + web search), crawl-rank (1-hop crawl from seed URL + embed + cosine rank), or auto.\",\n    },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n  },\n};\n\nexport const RESEARCH_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    question: { type: 'string', description: 'The research question to investigate' },\n    depth: {\n      type: 'string',\n      enum: ['quick', 'standard', 'comprehensive'],\n      description: 'Research depth: quick (~15s), standard (~40s, default), comprehensive (~80s)',\n    },\n    max_sources: {\n      type: 'number',\n      description: 'Override the default source count for the chosen depth (max 50)',\n    },\n    include_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Only search results from these domains',\n    },\n    exclude_domains: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Exclude results from these domains',\n    },\n    schema: {\n      type: 'object',\n      description: 'Optional JSON Schema -- structure the report to extract these fields',\n    },\n    stream: {\n      type: 'boolean',\n      description: 'Send progress notifications as each research phase completes',\n    },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n  },\n  required: ['question'],\n};\n\nexport const AGENT_TOOL_SCHEMA = {\n  type: 'object' as const,\n  properties: {\n    prompt: {\n      type: 'string',\n      description: 'Natural-language description of what data to gather',\n    },\n    urls: {\n      type: 'array',\n      items: { type: 'string' },\n      description: 'Specific URLs to include in the data gathering',\n    },\n    schema: {\n      type: 'object',\n      description: 'Optional JSON Schema -- extract structured data matching this schema from each page',\n    },\n    max_pages: {\n      type: 'number',\n      description: 'Maximum pages to fetch (default 10, max 100)',\n    },\n    max_time_ms: {\n      type: 'number',\n      description: 'Maximum execution time in milliseconds (default 60000)',\n    },\n    stream: {\n      type: 'boolean',\n      description: 'Send progress notifications as each step completes',\n    },\n    max_tokens_out: {\n      type: 'number',\n      description: \"Token-budget cap on total output. Uses cl100k-base BPE; non-OpenAI tokenizer counts may drift ~5-15%. When both max_tokens_out and max_chars are set, max_tokens_out wins.\",\n    },\n    include_full_markdown: {\n      type: 'boolean',\n      description: 'Include full markdown body in the response. Default false on multi-result tools (returns evidence excerpts only); set true to restore.',\n    },\n    citation_format: {\n      type: 'string',\n      enum: ['numbered', 'anthropic_tags', 'json'],\n      description: \"Citation rendering style. 'numbered' (default) inline [N] markers; 'json' returns a citations[] array; 'anthropic_tags' wraps sources in <source id='...'> tags.\",\n    },\n  },\n  required: ['prompt'],\n};\n\nexport const TOOL_SCHEMAS: Record<ToolName, ToolSchema> = {\n  fetch: FETCH_TOOL_SCHEMA,\n  search: SEARCH_TOOL_SCHEMA,\n  crawl: CRAWL_TOOL_SCHEMA,\n  cache: CACHE_TOOL_SCHEMA,\n  extract: EXTRACT_TOOL_SCHEMA,\n  find_similar: FIND_SIMILAR_TOOL_SCHEMA,\n  research: RESEARCH_TOOL_SCHEMA,\n  agent: AGENT_TOOL_SCHEMA,\n};\n"],"mappings":"AAQO,MAAM,oBAAoB;AAAA,EAC/B,MAAM;AAAA,EACN,YAAY;AAAA,IACV,KAAK,EAAE,MAAM,UAAU,aAAa,eAAe;AAAA,IACnD,WAAW;AAAA,MACT,MAAM;AAAA,MACN,MAAM,CAAC,QAAQ,UAAU,OAAO;AAAA,MAChC,aAAa;AAAA,IACf;AAAA,IACA,UAAU;AAAA,MACR,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,WAAW;AAAA,MACT,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,mBAAmB;AAAA,MACjB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,SAAS;AAAA,MACP,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,YAAY;AAAA,MACV,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,SAAS;AAAA,MACP,MAAM;AAAA,MACN,aAAa;AAAA,MACb,sBAAsB,EAAE,MAAM,SAAS;AAAA,IACzC;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,IACA,SAAS;AAAA,MACP,MAAM;AAAA,MACN,aACE;AAAA,MAEF,OAAO;AAAA,QACL,MAAM;AAAA,QACN,YAAY;AAAA,UACV,MAAM;AAAA,YACJ,MAAM;AAAA,YACN,MAAM,CAAC,SAAS,QAAQ,QAAQ,YAAY,UAAU,YAAY;AAAA,YAClE,aAAa;AAAA,UACf;AAAA,UACA,UAAU;AAAA,YACR,MAAM;AAAA,YACN,aAAa;AAAA,UACf;AAAA,UACA,MAAM;AAAA,YACJ,MAAM;AAAA,YACN,aAAa;AAAA,UACf;AAAA,UACA,IAAI;AAAA,YACF,MAAM;AAAA,YACN,aAAa;AAAA,UACf;AAAA,UACA,SAAS;AAAA,YACP,MAAM;AAAA,YACN,aAAa;AAAA,UACf;AAAA,UACA,WAAW;AAAA,YACT,MAAM;AAAA,YACN,MAAM,CAAC,QAAQ,IAAI;AAAA,YACnB,aAAa;AAAA,UACf;AAAA,UACA,QAAQ;AAAA,YACN,MAAM;AAAA,YACN,aAAa;AAAA,UACf;AAAA,QACF;AAAA,QACA,UAAU,CAAC,MAAM;AAAA,MACnB;AAAA,IACF;AAAA,IACA,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,MAAM,CAAC,SAAS,WAAW,SAAS;AAAA,MACpC,aAAa;AAAA,IACf;AAAA,EACF;AAAA,EACA,UAAU,CAAC,KAAK;AAClB;AAEO,MAAM,qBAAqB;AAAA,EAChC,MAAM;AAAA,EACN,YAAY;AAAA,IACV,OAAO;AAAA,MACL,OAAO;AAAA,QACL,EAAE,MAAM,UAAU,aAAa,eAAe;AAAA,QAC9C;AAAA,UACE,MAAM;AAAA,UACN,OAAO,EAAE,MAAM,SAAS;AAAA,UACxB,aAAa;AAAA,QACf;AAAA,MACF;AAAA,MACA,aAAa;AAAA,IACf;AAAA,IACA,aAAa,EAAE,MAAM,UAAU,aAAa,4CAA4C;AAAA,IACxF,aAAa,EAAE,MAAM,UAAU,aAAa,4LAA4L;AAAA,IACxO,iBAAiB,EAAE,MAAM,WAAW,aAAa,gDAAgD;AAAA,IACjG,mBAAmB,EAAE,MAAM,UAAU,aAAa,6DAA6D;AAAA,IAC/G,mBAAmB,EAAE,MAAM,UAAU,aAAa,wGAAwG;AAAA,IAC1J,iBAAiB,EAAE,MAAM,UAAU,aAAa,qDAAqD;AAAA,IACrG,YAAY,EAAE,MAAM,UAAU,MAAM,CAAC,OAAO,QAAQ,SAAS,MAAM,GAAG,aAAa,oBAAoB;AAAA,IACvG,gBAAgB,EAAE,MAAM,SAAS,OAAO,EAAE,MAAM,SAAS,GAAG,aAAa,4BAA4B;AAAA,IACrG,UAAU,EAAE,MAAM,UAAU,aAAa,sBAAsB;AAAA,IAC/D,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,WAAW;AAAA,MACT,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,SAAS;AAAA,MACP,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,UAAU;AAAA,MACR,MAAM;AAAA,MACN,MAAM,CAAC,WAAW,QAAQ,QAAQ,QAAQ,UAAU,QAAQ;AAAA,MAC5D,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,MAAM,CAAC,UAAU,eAAe;AAAA,MAChC,aACE;AAAA,IACJ;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,IACA,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,MAAM,CAAC,SAAS,WAAW,SAAS;AAAA,MACpC,aAAa;AAAA,IACf;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aACE;AAAA,MACF,YAAY;AAAA,QACV,MAAM,EAAE,MAAM,UAAU,aAAa,gDAAgD;AAAA,QACrF,aAAa,EAAE,MAAM,SAAS,OAAO,EAAE,MAAM,SAAS,GAAG,aAAa,mCAAmC;AAAA,QACzG,QAAQ,EAAE,MAAM,UAAU,aAAa,oDAAoD;AAAA,MAC7F;AAAA,MACA,sBAAsB;AAAA,IACxB;AAAA,EACF;AAAA,EACA,UAAU,CAAC,OAAO;AACpB;AAEO,MAAM,oBAAoB;AAAA,EAC/B,MAAM;AAAA,EACN,YAAY;AAAA,IACV,KAAK,EAAE,MAAM,UAAU,aAAa,kCAAkC;AAAA,IACtE,WAAW,EAAE,MAAM,UAAU,aAAa,4CAA4C;AAAA,IACtF,WAAW,EAAE,MAAM,UAAU,aAAa,uCAAuC;AAAA,IACjF,UAAU;AAAA,MACR,MAAM;AAAA,MACN,MAAM,CAAC,OAAO,OAAO,WAAW,KAAK;AAAA,MACrC,aAAa;AAAA,IACf;AAAA,IACA,kBAAkB;AAAA,MAChB,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,kBAAkB;AAAA,MAChB,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,UAAU,EAAE,MAAM,WAAW,aAAa,+CAA+C;AAAA,IACzF,eAAe,EAAE,MAAM,WAAW,aAAa,mDAAmD;AAAA,IAClG,iBAAiB,EAAE,MAAM,UAAU,aAAa,qDAAqD;AAAA,IACrG,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,EACF;AAAA,EACA,UAAU,CAAC,KAAK;AAClB;AAEO,MAAM,oBAAoB;AAAA,EAC/B,MAAM;AAAA,EACN,YAAY;AAAA,IACV,OAAO,EAAE,MAAM,UAAU,aAAa,uCAAuC;AAAA,IAC7E,aAAa;AAAA,MACX,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,OAAO;AAAA,MACL,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,OAAO;AAAA,MACL,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,OAAO;AAAA,MACL,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aACE;AAAA,IAGJ;AAAA,IACA,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,MAAM,CAAC,OAAO,QAAQ;AAAA,MACtB,aACE;AAAA,IAGJ;AAAA,IACA,OAAO;AAAA,MACL,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,EACF;AACF;AAEO,MAAM,sBAAsB;AAAA,EACjC,MAAM;AAAA,EACN,YAAY;AAAA,IACV,KAAK,EAAE,MAAM,UAAU,aAAa,gCAAgC;AAAA,IACpE,MAAM,EAAE,MAAM,UAAU,aAAa,iEAAiE;AAAA,IACtG,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,UAAU,YAAY,UAAU,YAAY;AAAA,MAC/D,aAAa;AAAA,IACf;AAAA,IACA,cAAc;AAAA,MACZ,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,UAAU;AAAA,MACR,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,cAAc;AAAA,MACZ,MAAM;AAAA,MACN,MAAM,CAAC,WAAW,UAAU,WAAW,eAAe,SAAS,cAAc;AAAA,MAC7E,aAAa;AAAA,IACf;AAAA,EACF;AACF;AAEO,MAAM,2BAA2B;AAAA,EACtC,MAAM;AAAA,EACN,YAAY;AAAA,IACV,KAAK;AAAA,MACH,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,SAAS;AAAA,MACP,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,aAAa;AAAA,MACX,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,eAAe;AAAA,MACb,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,aAAa;AAAA,MACX,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,MAAM,CAAC,QAAQ,SAAS,iBAAiB,YAAY;AAAA,MACrD,SAAS;AAAA,MACT,aAAa;AAAA,IACf;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,EACF;AACF;AAEO,MAAM,uBAAuB;AAAA,EAClC,MAAM;AAAA,EACN,YAAY;AAAA,IACV,UAAU,EAAE,MAAM,UAAU,aAAa,uCAAuC;AAAA,IAChF,OAAO;AAAA,MACL,MAAM;AAAA,MACN,MAAM,CAAC,SAAS,YAAY,eAAe;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,IACA,aAAa;AAAA,MACX,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,EACF;AAAA,EACA,UAAU,CAAC,UAAU;AACvB;AAEO,MAAM,oBAAoB;AAAA,EAC/B,MAAM;AAAA,EACN,YAAY;AAAA,IACV,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,MAAM;AAAA,MACJ,MAAM;AAAA,MACN,OAAO,EAAE,MAAM,SAAS;AAAA,MACxB,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,WAAW;AAAA,MACT,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,aAAa;AAAA,MACX,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,QAAQ;AAAA,MACN,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,gBAAgB;AAAA,MACd,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,uBAAuB;AAAA,MACrB,MAAM;AAAA,MACN,aAAa;AAAA,IACf;AAAA,IACA,iBAAiB;AAAA,MACf,MAAM;AAAA,MACN,MAAM,CAAC,YAAY,kBAAkB,MAAM;AAAA,MAC3C,aAAa;AAAA,IACf;AAAA,EACF;AAAA,EACA,UAAU,CAAC,QAAQ;AACrB;AAEO,MAAM,eAA6C;AAAA,EACxD,OAAO;AAAA,EACP,QAAQ;AAAA,EACR,OAAO;AAAA,EACP,OAAO;AAAA,EACP,SAAS;AAAA,EACT,cAAc;AAAA,EACd,UAAU;AAAA,EACV,OAAO;AACT;","names":[]}

package/dist/server/warmup-on-start.d.ts

@@ -0,0 +1,9 @@
+export declare function isEagerWarmupEnabled(): boolean;
+/**
+ * Fire-and-forget pre-warm of embed + rerank providers. No-op when env unset.
+ * Errors logged at warn, never thrown.
+ */
+export declare function maybeEagerWarmup(): void;
+/** Test-only: returns the in-flight warmup promise (or null). */
+export declare function _getWarmupPendingForTest(): Promise<void> | null;
+//# sourceMappingURL=warmup-on-start.d.ts.map

package/dist/server/warmup-on-start.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"warmup-on-start.d.ts","sourceRoot":"","sources":["../../src/server/warmup-on-start.ts"],"names":[],"mappings":"AAUA,wBAAgB,oBAAoB,IAAI,OAAO,CAE9C;AAED;;;GAGG;AACH,wBAAgB,gBAAgB,IAAI,IAAI,CAQvC;AAoCD,iEAAiE;AACjE,wBAAgB,wBAAwB,IAAI,OAAO,CAAC,IAAI,CAAC,GAAG,IAAI,CAE/D"}

package/dist/server/warmup-on-start.js

@@ -0,0 +1,55 @@
+import { createLogger } from "../logger.js";
+const log = createLogger("server");
+let pending = null;
+function isEagerWarmupEnabled() {
+  return process.env.WIGOLO_EAGER_WARMUP === "1";
+}
+function maybeEagerWarmup() {
+  if (!isEagerWarmupEnabled()) return;
+  pending = (async () => {
+    await Promise.all([warmEmbed(), warmRerank()]);
+  })().finally(() => {
+    pending = null;
+  });
+}
+async function warmEmbed() {
+  const start = Date.now();
+  try {
+    const mod = await import("../providers/embed-provider.js");
+    const provider = await mod.getEmbedProvider();
+    if (typeof provider.warmup === "function") {
+      await provider.warmup();
+    }
+    log.info("eager warmup: embed ready", { ms: Date.now() - start });
+  } catch (err) {
+    log.warn("eager warmup: embed failed", {
+      ms: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err)
+    });
+  }
+}
+async function warmRerank() {
+  const start = Date.now();
+  try {
+    const mod = await import("../providers/rerank-provider.js");
+    const provider = await mod.getRerankProvider();
+    if (typeof provider.warmup === "function") {
+      await provider.warmup();
+    }
+    log.info("eager warmup: rerank ready", { ms: Date.now() - start });
+  } catch (err) {
+    log.warn("eager warmup: rerank failed", {
+      ms: Date.now() - start,
+      error: err instanceof Error ? err.message : String(err)
+    });
+  }
+}
+function _getWarmupPendingForTest() {
+  return pending;
+}
+export {
+  _getWarmupPendingForTest,
+  isEagerWarmupEnabled,
+  maybeEagerWarmup
+};
+//# sourceMappingURL=warmup-on-start.js.map

package/dist/server/warmup-on-start.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../src/server/warmup-on-start.ts"],"sourcesContent":["import { createLogger } from '../logger.js';\n\nconst log = createLogger('server');\n\ninterface MaybeWarmable {\n  warmup?: () => Promise<void>;\n}\n\nlet pending: Promise<void> | null = null;\n\nexport function isEagerWarmupEnabled(): boolean {\n  return process.env.WIGOLO_EAGER_WARMUP === '1';\n}\n\n/**\n * Fire-and-forget pre-warm of embed + rerank providers. No-op when env unset.\n * Errors logged at warn, never thrown.\n */\nexport function maybeEagerWarmup(): void {\n  if (!isEagerWarmupEnabled()) return;\n\n  pending = (async () => {\n    await Promise.all([warmEmbed(), warmRerank()]);\n  })().finally(() => {\n    pending = null;\n  });\n}\n\nasync function warmEmbed(): Promise<void> {\n  const start = Date.now();\n  try {\n    const mod = await import('../providers/embed-provider.js');\n    const provider = (await mod.getEmbedProvider()) as MaybeWarmable;\n    if (typeof provider.warmup === 'function') {\n      await provider.warmup();\n    }\n    log.info('eager warmup: embed ready', { ms: Date.now() - start });\n  } catch (err) {\n    log.warn('eager warmup: embed failed', {\n      ms: Date.now() - start,\n      error: err instanceof Error ? err.message : String(err),\n    });\n  }\n}\n\nasync function warmRerank(): Promise<void> {\n  const start = Date.now();\n  try {\n    const mod = await import('../providers/rerank-provider.js');\n    const provider = (await mod.getRerankProvider()) as MaybeWarmable;\n    if (typeof provider.warmup === 'function') {\n      await provider.warmup();\n    }\n    log.info('eager warmup: rerank ready', { ms: Date.now() - start });\n  } catch (err) {\n    log.warn('eager warmup: rerank failed', {\n      ms: Date.now() - start,\n      error: err instanceof Error ? err.message : String(err),\n    });\n  }\n}\n\n/** Test-only: returns the in-flight warmup promise (or null). */\nexport function _getWarmupPendingForTest(): Promise<void> | null {\n  return pending;\n}\n"],"mappings":"AAAA,SAAS,oBAAoB;AAE7B,MAAM,MAAM,aAAa,QAAQ;AAMjC,IAAI,UAAgC;AAE7B,SAAS,uBAAgC;AAC9C,SAAO,QAAQ,IAAI,wBAAwB;AAC7C;AAMO,SAAS,mBAAyB;AACvC,MAAI,CAAC,qBAAqB,EAAG;AAE7B,aAAW,YAAY;AACrB,UAAM,QAAQ,IAAI,CAAC,UAAU,GAAG,WAAW,CAAC,CAAC;AAAA,EAC/C,GAAG,EAAE,QAAQ,MAAM;AACjB,cAAU;AAAA,EACZ,CAAC;AACH;AAEA,eAAe,YAA2B;AACxC,QAAM,QAAQ,KAAK,IAAI;AACvB,MAAI;AACF,UAAM,MAAM,MAAM,OAAO,gCAAgC;AACzD,UAAM,WAAY,MAAM,IAAI,iBAAiB;AAC7C,QAAI,OAAO,SAAS,WAAW,YAAY;AACzC,YAAM,SAAS,OAAO;AAAA,IACxB;AACA,QAAI,KAAK,6BAA6B,EAAE,IAAI,KAAK,IAAI,IAAI,MAAM,CAAC;AAAA,EAClE,SAAS,KAAK;AACZ,QAAI,KAAK,8BAA8B;AAAA,MACrC,IAAI,KAAK,IAAI,IAAI;AAAA,MACjB,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAAA,IACxD,CAAC;AAAA,EACH;AACF;AAEA,eAAe,aAA4B;AACzC,QAAM,QAAQ,KAAK,IAAI;AACvB,MAAI;AACF,UAAM,MAAM,MAAM,OAAO,iCAAiC;AAC1D,UAAM,WAAY,MAAM,IAAI,kBAAkB;AAC9C,QAAI,OAAO,SAAS,WAAW,YAAY;AACzC,YAAM,SAAS,OAAO;AAAA,IACxB;AACA,QAAI,KAAK,8BAA8B,EAAE,IAAI,KAAK,IAAI,IAAI,MAAM,CAAC;AAAA,EACnE,SAAS,KAAK;AACZ,QAAI,KAAK,+BAA+B;AAAA,MACtC,IAAI,KAAK,IAAI,IAAI;AAAA,MACjB,OAAO,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG;AAAA,IACxD,CAAC;AAAA,EACH;AACF;AAGO,SAAS,2BAAiD;AAC/D,SAAO;AACT;","names":[]}

package/dist/server.d.ts

@@ -1,2 +1,19 @@
+import { Server } from '@modelcontextprotocol/sdk/server/index.js';
+import { SmartRouter } from './fetch/router.js';
+import { MultiBrowserPool } from './fetch/browser-pool.js';
+import { BackendStatus } from './server/backend-status.js';
+import { PluginRegistry } from './plugins/registry.js';
+import type { SearchEngine } from './types.js';
+export interface Subsystems {
+    searchEngines: SearchEngine[];
+    browserPool: MultiBrowserPool;
+    router: SmartRouter;
+    backendStatus: BackendStatus;
+    pluginRegistry: PluginRegistry;
+    shutdown: () => Promise<void>;
+    bootstrapSearxng: () => Promise<void>;
+}
+export declare function initSubsystems(): Promise<Subsystems>;
+export declare function createMcpServer(subsystems: Subsystems): Server;
 export declare function startServer(): Promise<void>;
 //# sourceMappingURL=server.d.ts.map

package/dist/server.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"server.d.ts","sourceRoot":"","sources":["../src/server.ts"],"names":[],"mappings":"AAsLA,wBAAsB,WAAW,IAAI,OAAO,CAAC,IAAI,CAAC,CA6JjD"}
+{"version":3,"file":"server.d.ts","sourceRoot":"","sources":["../src/server.ts"],"names":[],"mappings":"AAGA,OAAO,EAAE,MAAM,EAAE,MAAM,2CAA2C,CAAC;AAQnE,OAAO,EAAE,WAAW,EAAmB,MAAM,mBAAmB,CAAC;AACjE,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAC;AAoB3D,OAAO,EAAE,aAAa,EAAE,MAAM,4BAA4B,CAAC;AAsB3D,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAEvD,OAAO,KAAK,EAA2B,YAAY,EAAuG,MAAM,YAAY,CAAC;AAkB7K,MAAM,WAAW,UAAU;IACzB,aAAa,EAAE,YAAY,EAAE,CAAC;IAC9B,WAAW,EAAE,gBAAgB,CAAC;IAC9B,MAAM,EAAE,WAAW,CAAC;IACpB,aAAa,EAAE,aAAa,CAAC;IAC7B,cAAc,EAAE,cAAc,CAAC;IAC/B,QAAQ,EAAE,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC;IAC9B,gBAAgB,EAAE,MAAM,OAAO,CAAC,IAAI,CAAC,CAAC;CACvC;AAED,wBAAsB,cAAc,IAAI,OAAO,CAAC,UAAU,CAAC,CAuL1D;AAED,wBAAgB,eAAe,CAAC,UAAU,EAAE,UAAU,GAAG,MAAM,CAwO9D;AAED,wBAAsB,WAAW,IAAI,OAAO,CAAC,IAAI,CAAC,CAsBjD"}