Pricing Page Scan

Instructions

Implementation Reference

• src/index.ts:765-844 (registration)

  The check_pricing tool is registered via this.server.registerTool() with the name 'check_pricing'. This is the registration and handler combined — the tool is defined with inputSchema, outputSchema, and the async handler function all within the registerTool call.

  // ───────────────────────────────────────────────
  // PAID $0.02: check_pricing
  // ───────────────────────────────────────────────
  this.server.registerTool(
    "check_pricing",
    {
      title: "Pricing Page Scan",
      description:
        "Fetch a public pricing page and extract first-pass pricing signals before " +
        "you quote plan costs, free tiers, or plan names. Use this when you already " +
        "have a likely pricing URL and need a quick live scan of visible page text. " +
        "It returns price-like strings, heuristic plan labels, free or free-trial " +
        "signals, and cache information. It does not map prices to exact plans, " +
        "normalize currencies, execute checkout flows, or guarantee that a price " +
        "applies to a specific region or customer type. JavaScript-rendered, " +
        "logged-in, or heavily obfuscated pricing details can be missed. Results " +
        "are cached for 5 minutes.",
      inputSchema: {
        url: z.string().url().describe(
          "Public pricing or plans URL to analyze. Prefer the specific pricing page, for example https://stripe.com/pricing, rather than a generic homepage.",
        ),
      },
      outputSchema: {
        url: z.string().describe(
          "Pricing page that was analyzed.",
        ),
        cached: z.boolean().describe(
          "True when the page body came from the 5-minute cache instead of a new fetch.",
        ).optional(),
        pricesFound: z.array(z.string()).describe(
          "Distinct price-like strings extracted from the page text. These are not linked back to specific plans or billing conditions.",
        ).optional(),
        plansDetected: z.array(z.string()).describe(
          "Lowercased heuristic plan labels detected from the page text. They are useful hints, not authoritative plan identifiers.",
        ).optional(),
        hasFreeOption: z.boolean().describe(
          "True when the page contains signals that a free plan or $0 option exists somewhere on the page. This is a page-level signal, not proof that the offer is currently self-serve or globally available.",
        ).optional(),
        hasFreeTrial: z.boolean().describe(
          "True when the page contains signals that a free trial exists somewhere on the page.",
        ).optional(),
        pageLength: z.number().int().nonnegative().describe(
          "Size of the fetched page body in characters.",
        ).optional(),
        error: z.string().describe(
          "Fetch or parsing error when the pricing page could not be analyzed.",
        ).optional(),
      },
      annotations: readOnlyNetworkToolAnnotations,
    },
    async ({ url }) => {
      try {
        const { body, fromCache } = await cachedFetch(sql, url);
        // Extract pricing signals from page content
        const priceRegex = /\$\d[\d,]*(?:\.\d{2})?(?:\s*\/\s*(?:mo(?:nth)?|yr|year|user|seat|req|call|token))?/gi;
        const prices = [...new Set(body.match(priceRegex) || [])].slice(0, 20);
        const planRegex = /(?:free|starter|basic|pro|premium|enterprise|business|team|hobby|growth|scale)\s*(?:plan|tier)?/gi;
        const plans = [...new Set((body.match(planRegex) || []).map(p => p.trim().toLowerCase()))];
        const hasFree = /free\s*(?:plan|tier|forever|trial)|(?:\$0|0\.00)/i.test(body);
        const hasFreeTrial = /free\s*trial|try\s*(?:it\s*)?free|start\s*free/i.test(body);
  
        logUsage("check_pricing", true);
        return structuredToolResult({
          url,
          cached: fromCache,
          pricesFound: prices,
          plansDetected: plans,
          hasFreeOption: hasFree,
          hasFreeTrial: hasFreeTrial,
          pageLength: body.length,
        });
      } catch (e: unknown) {
        logUsage("check_pricing", false);
        return structuredToolResult({
          url,
          error: e instanceof Error ? e.message : String(e),
        });
      }
    }
  );

• src/index.ts:786-813 (schema)

  Input and output schema for check_pricing. Input: url (string). Output: url, cached, pricesFound, plansDetected, hasFreeOption, hasFreeTrial, pageLength, error.

  },
  outputSchema: {
    url: z.string().describe(
      "Pricing page that was analyzed.",
    ),
    cached: z.boolean().describe(
      "True when the page body came from the 5-minute cache instead of a new fetch.",
    ).optional(),
    pricesFound: z.array(z.string()).describe(
      "Distinct price-like strings extracted from the page text. These are not linked back to specific plans or billing conditions.",
    ).optional(),
    plansDetected: z.array(z.string()).describe(
      "Lowercased heuristic plan labels detected from the page text. They are useful hints, not authoritative plan identifiers.",
    ).optional(),
    hasFreeOption: z.boolean().describe(
      "True when the page contains signals that a free plan or $0 option exists somewhere on the page. This is a page-level signal, not proof that the offer is currently self-serve or globally available.",
    ).optional(),
    hasFreeTrial: z.boolean().describe(
      "True when the page contains signals that a free trial exists somewhere on the page.",
    ).optional(),
    pageLength: z.number().int().nonnegative().describe(
      "Size of the fetched page body in characters.",
    ).optional(),
    error: z.string().describe(
      "Fetch or parsing error when the pricing page could not be analyzed.",
    ).optional(),
  },
  annotations: readOnlyNetworkToolAnnotations,

• src/index.ts:815-843 (handler)

  The async handler function for check_pricing. It fetches the pricing URL, extracts price strings via regex, detects plan labels, checks for free/trial signals, and returns the structured result.

  async ({ url }) => {
    try {
      const { body, fromCache } = await cachedFetch(sql, url);
      // Extract pricing signals from page content
      const priceRegex = /\$\d[\d,]*(?:\.\d{2})?(?:\s*\/\s*(?:mo(?:nth)?|yr|year|user|seat|req|call|token))?/gi;
      const prices = [...new Set(body.match(priceRegex) || [])].slice(0, 20);
      const planRegex = /(?:free|starter|basic|pro|premium|enterprise|business|team|hobby|growth|scale)\s*(?:plan|tier)?/gi;
      const plans = [...new Set((body.match(planRegex) || []).map(p => p.trim().toLowerCase()))];
      const hasFree = /free\s*(?:plan|tier|forever|trial)|(?:\$0|0\.00)/i.test(body);
      const hasFreeTrial = /free\s*trial|try\s*(?:it\s*)?free|start\s*free/i.test(body);
  
      logUsage("check_pricing", true);
      return structuredToolResult({
        url,
        cached: fromCache,
        pricesFound: prices,
        plansDetected: plans,
        hasFreeOption: hasFree,
        hasFreeTrial: hasFreeTrial,
        pageLength: body.length,
      });
    } catch (e: unknown) {
      logUsage("check_pricing", false);
      return structuredToolResult({
        url,
        error: e instanceof Error ? e.message : String(e),
      });
    }
  }

• src/index.ts:491-498 (helper)

  cachedFetch helper used by check_pricing to fetch and cache the pricing page body with a 5-minute TTL.

  async function cachedFetch(sql: SqlTagFn, url: string): Promise<{ body: string; fromCache: boolean }> {
    const cached = cacheGet(sql, url);
    if (cached) return { body: cached, fromCache: true };
    const resp = await fetch(url, { headers: { "User-Agent": "GroundTruth/0.3" } });
    const body = await resp.text();
    if (resp.ok) cacheSet(sql, url, body);
    return { body, fromCache: false };
  }

• src/index.ts:467-488 (helper)

  cacheGet and cacheSet helpers used by cachedFetch for SQLite-backed caching of fetched page bodies.

  // --- Cache helpers using Durable Object SQLite tagged template ---
  function cacheGet(sql: SqlTagFn, key: string): string | null {
    const rows = sql<{ data: string; ts: number }>`SELECT data, ts FROM cache WHERE key = ${key}`;
    if (rows.length === 0) return null;
    const row = rows[0];
    if (Date.now() - row.ts > CACHE_TTL_MS) {
      sql`DELETE FROM cache WHERE key = ${key}`;
      return null;
    }
    return row.data;
  }
  
  function cacheSet(sql: SqlTagFn, key: string, data: string): void {
    if (data.length > MAX_CACHEABLE_BODY_BYTES) return;
  
    const ts = Date.now();
    try {
      sql`INSERT OR REPLACE INTO cache (key, data, ts) VALUES (${key}, ${data}, ${ts})`;
    } catch {
      // Cache writes should never take down a verification request.
    }
  }