india-reg-mcp 1.0.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Akhil Govind
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -0,0 +1,292 @@
+# india-reg-mcp
+
+An MCP server that gives Claude (and any MCP client) searchable, cited access to **RBI and SEBI regulatory documents** — circulars, master directions, notifications, and regulations.
+
+No API keys. No subscriptions. Built entirely on free public government data.
+
+---
+
+## Why This Exists
+
+Regulatory rules in India are scattered across thousands of PDFs and HTML pages on rbi.org.in and sebi.gov.in. They reference and supersede each other constantly. There is no AI-accessible way to ask "what are the current rules on X" and get sourced answers.
+
+This MCP solves that. It scrapes, indexes, and exposes the full text of RBI and SEBI documents in a local SQLite database — so Claude can search and retrieve primary-source regulatory text with official citations, instantly.
+
+**This is not a legal advice tool.** It retrieves primary-source documents so you can reason over actual rules instead of hallucinated summaries.
+
+---
+
+## What It Does
+
+- **Full-text search** across all indexed RBI and SEBI documents
+- **Retrieve full document body** by ID, with official source link
+- **List recent documents** — useful for "what changed this month" questions
+- **List Master Directions and Master Circulars** — the consolidated, currently-in-force rules on each subject
+- **Browse by department** — SEBI's Investment Management, Market Regulation, etc.
+- **On-demand sync** — pull newly published documents from regulators' sites
+- **Topic search** — returns both consolidated master rules AND recent circulars on a subject in one call
+
+---
+
+## Tools (8 total)
+
+| Tool | Description |
+|---|---|
+| `search_regulations` | Full-text search with optional regulator/type filter |
+| `get_document` | Retrieve full text of a document by ID |
+| `get_recent` | Most recent documents, optionally filtered |
+| `list_master_directions` | All RBI Master Directions + SEBI Master Circulars |
+| `list_by_department` | SEBI documents by department name |
+| `sync_latest` | Incremental scrape of new documents from RBI + SEBI |
+| `sync_status` | Document count breakdown + last sync time |
+| `search_by_topic` | Combined master rules + recent circulars on a topic |
+
+---
+
+## Architecture
+
+```
+RBI / SEBI websites
+        │
+        ▼
+   Scrapers (TypeScript)
+   rbi.ts  ──── ASP.NET POST with viewstate tokens
+   sebi.ts ──── AJAX pagination via JSP endpoint
+        │
+        ▼
+   SQLite DB  (~/.india-reg-mcp/regdata.db)
+   ├── documents table (full text, metadata, source URLs)
+   ├── documents_fts (FTS5 full-text index, porter stemmer)
+   └── sync_meta (last sync timestamp)
+        │
+        ▼
+   MCP Server (stdio transport)
+   ├── 8 tools exposed to Claude
+   └── Every result includes official source URL + disclaimer
+```
+
+**Key design decision:** Tools never scrape live. The scrapers populate a local SQLite database once, and tools query that instantly. Only `sync_latest` hits the regulators' sites. This makes every tool call fast, keeps you off government servers during normal use, and works offline once indexed.
+
+---
+
+## Installation
+
+### Prerequisites
+
+- Node.js 22 LTS or higher
+- Claude Desktop or Claude Code CLI
+
+### Setup
+
+```bash
+git clone https://github.com/Akhilgovind02/india-regulatory-mcp.git
+cd india-regulatory-mcp
+npm install
+npm run build
+```
+
+### First-run sync (populates the database)
+
+```bash
+npm run sync
+```
+
+This scrapes RBI notifications for the last 36 months and SEBI circulars (~1000 recent documents). Takes **5–15 minutes** depending on your connection. The database is stored at `~/.india-reg-mcp/regdata.db` and survives rebuilds.
+
+Progress is printed to stderr:
+```
+[sync] RBI 2026-6: 12 docs found
+[sync] RBI 2026-5: 18 docs found
+...
+[sync] SEBI ssid=6 page 0: 25 docs
+...
+[sync] Sync complete.
+```
+
+---
+
+## Configuration
+
+### Claude Code (CLI)
+
+```bash
+claude mcp add -s user india-reg node /ABSOLUTE/PATH/TO/india-regulatory-mcp/dist/index.js
+```
+
+Or add to `~/.claude.json` manually:
+```json
+{
+  "mcpServers": {
+    "india-reg": {
+      "command": "node",
+      "args": ["/ABSOLUTE/PATH/TO/india-regulatory-mcp/dist/index.js"]
+    }
+  }
+}
+```
+
+### Claude Desktop
+
+Edit `%APPDATA%\Claude\claude_desktop_config.json` (Windows) or `~/Library/Application Support/Claude/claude_desktop_config.json` (macOS):
+
+```json
+{
+  "mcpServers": {
+    "india-reg": {
+      "command": "node",
+      "args": ["C:\\ABSOLUTE\\PATH\\TO\\india-regulatory-mcp\\dist\\index.js"]
+    }
+  }
+}
+```
+
+Restart Claude Desktop fully after editing the config.
+
+---
+
+## Usage Examples
+
+Once connected, ask Claude naturally:
+
+```
+What are the current RBI rules on digital lending?
+```
+→ Claude calls `search_by_topic("digital lending")` — returns the Master Direction on Digital Lending plus recent amending circulars, each with official source links.
+
+```
+What new SEBI circulars came out this month?
+```
+→ Claude calls `get_recent(regulator="SEBI", limit=10)`.
+
+```
+Show me all RBI Master Directions
+```
+→ Claude calls `list_master_directions(regulator="RBI")`.
+
+```
+What are the KYC requirements for mutual funds?
+```
+→ Claude calls `search_regulations("KYC mutual fund", doc_type="master_direction")`.
+
+```
+Pull in the latest regulatory updates
+```
+→ Claude calls `sync_latest()` — incremental, only fetches new documents.
+
+---
+
+## Data Sources
+
+### RBI (Reserve Bank of India)
+- **Notifications + Circulars**: `rbi.org.in/Scripts/NotificationUser.aspx`
+- **Master Directions**: `rbi.org.in/Scripts/BS_ViewMasterDirections.aspx`
+- **Master Circulars**: `rbi.org.in/Scripts/BS_ViewMasterCirculardetails.aspx`
+- Document bodies: HTML page text (with PDF fallback for PDF-only docs)
+
+### SEBI (Securities and Exchange Board of India)
+- **Circulars** (`ssid=7`): ~2,775 documents as of June 2026
+- **Master Circulars** (`ssid=6`): consolidated current rules by topic
+- **Regulations** (`ssid=3`): statutory regulations
+- Document bodies: primarily PDF-embedded content extracted via `pdf-parse`
+
+---
+
+## Document Types
+
+| Type | Source | Description |
+|---|---|---|
+| `circular` | RBI + SEBI | Point-in-time regulatory guidance |
+| `master_direction` | RBI | Consolidated current rules on a subject (supersedes earlier circulars) |
+| `master_circular` | SEBI | Same as master direction, SEBI's term |
+| `notification` | RBI | Statutory notifications under various Acts |
+| `regulation` | SEBI | Formal regulations (e.g. SEBI (FPI) Regulations 2019) |
+
+**For compliance questions, start with `master_direction`/`master_circular`.** These represent the current state of rules, not a point-in-time snapshot.
+
+---
+
+## Keeping the Index Fresh
+
+The database is a point-in-time snapshot. RBI and SEBI publish new documents frequently (several per week).
+
+**Option 1 — Ask Claude:** "Pull in the latest regulatory updates" → Claude calls `sync_latest`.
+
+**Option 2 — CLI:**
+```bash
+npm run sync
+```
+
+**Option 3 — Scheduled (example cron, runs every Sunday at 2am):**
+```
+0 2 * * 0 cd /path/to/india-regulatory-mcp && npm run sync >> ~/.india-reg-mcp/sync.log 2>&1
+```
+
+`sync_latest` and `npm run sync` are both incremental — they skip documents already in the database.
+
+---
+
+## Project Structure
+
+```
+india-regulatory-mcp/
+├── src/
+│   ├── index.ts                  ← MCP server, all 8 tools
+│   ├── db/
+│   │   ├── schema.ts             ← SQLite schema + FTS5 setup
+│   │   └── queries.ts            ← DB read/write functions
+│   ├── scrapers/
+│   │   ├── rbi.ts                ← RBI scraper (ASP.NET POST + viewstate)
+│   │   ├── sebi.ts               ← SEBI scraper (AJAX pagination)
+│   │   ├── pdf.ts                ← PDF download + text extraction
+│   │   ├── run-sync.ts           ← CLI full sync runner
+│   │   └── repair-bodies.ts      ← Utility: backfill missing body text
+│   └── util/
+│       ├── http.ts               ← Polite fetch (UA + retry + delay)
+│       └── format.ts             ← MCP response helpers
+├── dist/                         ← Compiled output (not in repo)
+├── package.json
+└── tsconfig.json
+```
+
+---
+
+## Technical Notes
+
+### Scraping approach
+
+**RBI** uses ASP.NET with viewstate tokens. The scraper fetches viewstate on startup, then POSTs month-by-month to get document listings. Content is extracted from the HTML circular page; PDF fallback for PDF-only documents.
+
+**SEBI** uses a JSP-based listing with AJAX pagination. Page 1 is a GET request (establishes JSESSIONID). Pages 2+ POST to `getnewslistinfo.jsp` with the session cookie. Circular content is typically a PDF embedded in an iframe — the scraper extracts the PDF URL and parses it with `pdf-parse`.
+
+### Polite scraping
+- Concurrency capped at 2 parallel requests per host
+- 300ms delay between document fetches
+- 500ms delay between listing pages
+- Exponential backoff on 429/5xx responses
+- Real User-Agent string identifying the tool
+
+### Database
+- Location: `~/.india-reg-mcp/regdata.db`
+- WAL mode for performance
+- FTS5 with Porter stemmer — handles morphological variants ("lending" matches "lend", "lender")
+- Full-text search terms are quoted per-word to handle hyphens and special characters safely
+
+---
+
+## Stack
+
+| Package | Version | Purpose |
+|---|---|---|
+| `@modelcontextprotocol/sdk` | 1.29.0 | MCP server framework |
+| `better-sqlite3` | 12.10.0 | SQLite with built-in FTS5 |
+| `cheerio` | 1.2.0 | HTML parsing |
+| `pdf-parse` | 2.4.5 | PDF text extraction |
+| `turndown` | 7.2.4 | HTML → Markdown |
+| `p-limit` | 7.3.0 | Concurrency control |
+| `zod` | 4.4.3 | Schema validation |
+
+---
+
+## Disclaimer
+
+This tool retrieves and surfaces primary-source regulatory documents from official RBI and SEBI publications. It does not provide legal advice. Always verify against the official linked document. The index is a point-in-time snapshot — use `sync_latest` to pull recent publications.

package/dist/db/queries.js

@@ -0,0 +1,91 @@
+import { db } from "./schema.js";
+// Lazy statement cache — prepared after initSchema() has created the tables
+const stmts = {};
+function stmt(key, sql) {
+    if (!stmts[key])
+        stmts[key] = db.prepare(sql);
+    return stmts[key];
+}
+export function upsertDoc(doc) {
+    stmt("upsert", `
+    INSERT INTO documents (id, regulator, doc_type, title, date, department, source_url, pdf_url, body, indexed_at)
+    VALUES (@id, @regulator, @doc_type, @title, @date, @department, @source_url, @pdf_url, @body, @indexed_at)
+    ON CONFLICT(id) DO UPDATE SET
+      title=@title, body=@body, department=@department, pdf_url=@pdf_url, indexed_at=@indexed_at
+  `).run(doc);
+}
+export function upsertMany(docs) {
+    const s = stmt("upsert", `
+    INSERT INTO documents (id, regulator, doc_type, title, date, department, source_url, pdf_url, body, indexed_at)
+    VALUES (@id, @regulator, @doc_type, @title, @date, @department, @source_url, @pdf_url, @body, @indexed_at)
+    ON CONFLICT(id) DO UPDATE SET
+      title=@title, body=@body, department=@department, pdf_url=@pdf_url, indexed_at=@indexed_at
+  `);
+    const tx = db.transaction((rows) => rows.forEach((r) => s.run(r)));
+    tx(docs);
+}
+export function docExists(id) {
+    return !!stmt("exists", "SELECT 1 FROM documents WHERE id = ?").get(id);
+}
+export function getDoc(id) {
+    return stmt("getDoc", "SELECT * FROM documents WHERE id = ?").get(id);
+}
+export function searchDocs(opts) {
+    if (!opts.query.trim())
+        return [];
+    const limit = opts.limit ?? 10;
+    let sql = `
+    SELECT d.*, snippet(documents_fts, 1, '<<', '>>', ' … ', 16) AS snippet
+    FROM documents_fts f
+    JOIN documents d ON d.rowid = f.rowid
+    WHERE documents_fts MATCH @q
+  `;
+    const params = { q: escapeFts(opts.query) };
+    if (opts.regulator) {
+        sql += " AND d.regulator = @regulator";
+        params.regulator = opts.regulator;
+    }
+    if (opts.docType) {
+        sql += " AND d.doc_type = @docType";
+        params.docType = opts.docType;
+    }
+    sql += " ORDER BY rank, d.date DESC LIMIT @limit";
+    params.limit = limit;
+    return db.prepare(sql).all(params);
+}
+export function recentDocs(opts) {
+    const limit = opts.limit ?? 15;
+    let sql = "SELECT * FROM documents WHERE 1=1";
+    const params = {};
+    if (opts.regulator) {
+        sql += " AND regulator = @regulator";
+        params.regulator = opts.regulator;
+    }
+    if (opts.docType) {
+        sql += " AND doc_type = @docType";
+        params.docType = opts.docType;
+    }
+    sql += " ORDER BY date DESC LIMIT @limit";
+    params.limit = limit;
+    return db.prepare(sql).all(params);
+}
+export function listByDepartment(dept, limit = 20) {
+    return db.prepare("SELECT * FROM documents WHERE department LIKE ? ORDER BY date DESC LIMIT ?").all(`%${dept}%`, limit);
+}
+export function docCount() {
+    return db.prepare("SELECT regulator, doc_type, COUNT(*) as n FROM documents GROUP BY regulator, doc_type").all();
+}
+export function getSyncMeta(key) {
+    const row = stmt("getMeta", "SELECT value FROM sync_meta WHERE key = ?").get(key);
+    return row?.value;
+}
+export function setSyncMeta(key, value) {
+    stmt("setMeta", "INSERT INTO sync_meta (key,value) VALUES (?,?) ON CONFLICT(key) DO UPDATE SET value=?")
+        .run(key, value, value);
+}
+function escapeFts(q) {
+    const trimmed = q.trim();
+    if (!trimmed)
+        return "";
+    return trimmed.split(/\s+/).map((w) => `"${w.replace(/"/g, '')}"`).join(" ");
+}

package/dist/db/schema.js

@@ -0,0 +1,56 @@
+import Database from "better-sqlite3";
+import { homedir } from "node:os";
+import { join } from "node:path";
+import { mkdirSync, existsSync } from "node:fs";
+const DB_DIR = join(homedir(), ".india-reg-mcp");
+if (!existsSync(DB_DIR))
+    mkdirSync(DB_DIR, { recursive: true });
+const DB_PATH = join(DB_DIR, "regdata.db");
+export const db = new Database(DB_PATH);
+db.pragma("journal_mode = WAL");
+export function initSchema() {
+    db.exec(`
+    CREATE TABLE IF NOT EXISTS documents (
+      id           TEXT PRIMARY KEY,
+      regulator    TEXT NOT NULL,
+      doc_type     TEXT NOT NULL,
+      title        TEXT NOT NULL,
+      date         TEXT NOT NULL,
+      department   TEXT,
+      source_url   TEXT NOT NULL,
+      pdf_url      TEXT,
+      body         TEXT,
+      indexed_at   TEXT NOT NULL
+    );
+
+    CREATE INDEX IF NOT EXISTS idx_regulator ON documents(regulator);
+    CREATE INDEX IF NOT EXISTS idx_doctype   ON documents(doc_type);
+    CREATE INDEX IF NOT EXISTS idx_date      ON documents(date);
+
+    CREATE VIRTUAL TABLE IF NOT EXISTS documents_fts USING fts5(
+      title, body,
+      content='documents',
+      content_rowid='rowid',
+      tokenize='porter unicode61'
+    );
+
+    DROP TRIGGER IF EXISTS documents_ai;
+
+    CREATE TRIGGER documents_ai AFTER INSERT ON documents BEGIN
+      INSERT INTO documents_fts(documents_fts, rowid, title, body) VALUES ('delete', new.rowid, new.title, new.body);
+      INSERT INTO documents_fts(rowid, title, body) VALUES (new.rowid, new.title, new.body);
+    END;
+    CREATE TRIGGER IF NOT EXISTS documents_ad AFTER DELETE ON documents BEGIN
+      INSERT INTO documents_fts(documents_fts, rowid, title, body) VALUES ('delete', old.rowid, old.title, old.body);
+    END;
+    CREATE TRIGGER IF NOT EXISTS documents_au AFTER UPDATE ON documents BEGIN
+      INSERT INTO documents_fts(documents_fts, rowid, title, body) VALUES ('delete', old.rowid, old.title, old.body);
+      INSERT INTO documents_fts(rowid, title, body) VALUES (new.rowid, new.title, new.body);
+    END;
+
+    CREATE TABLE IF NOT EXISTS sync_meta (
+      key   TEXT PRIMARY KEY,
+      value TEXT
+    );
+  `);
+}

package/dist/index.js

@@ -0,0 +1,169 @@
+#!/usr/bin/env node
+import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
+import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
+import { z } from "zod";
+import { initSchema } from "./db/schema.js";
+import * as q from "./db/queries.js";
+import { syncRbi } from "./scrapers/rbi.js";
+import { syncSebi } from "./scrapers/sebi.js";
+import { DISCLAIMER, ok, err, emptyDbMsg } from "./util/format.js";
+initSchema();
+const server = new McpServer({ name: "india-reg-mcp", version: "1.0.0" });
+server.tool("search_regulations", "Full-text search across indexed RBI and SEBI regulatory documents (circulars, master directions, notifications, regulations). " +
+    "Returns matching documents with title, date, regulator, a highlighted snippet, and the official source link. " +
+    "Use this to answer 'what are the rules on X' style questions with cited primary sources.", {
+    query: z.string().describe("Search terms e.g. 'digital lending', 'KYC periodic updation', 'FPI registration', 'mutual fund nomination'"),
+    regulator: z.enum(["RBI", "SEBI"]).optional().describe("Optionally limit to one regulator"),
+    doc_type: z.enum(["circular", "master_direction", "master_circular", "notification", "regulation"]).optional()
+        .describe("Optionally limit to one document type. master_direction/master_circular are consolidated current rules."),
+    limit: z.number().default(10).describe("Max results (1-25)"),
+}, async ({ query, regulator, doc_type, limit }) => {
+    try {
+        if (!q.docCount().length)
+            return emptyDbMsg();
+        const results = q.searchDocs({ query, regulator, docType: doc_type, limit: Math.min(limit, 25) });
+        return ok({
+            query,
+            resultCount: results.length,
+            results: results.map((r) => ({
+                id: r.id, regulator: r.regulator, type: r.doc_type,
+                title: r.title, date: r.date, snippet: r.snippet,
+                source: r.source_url, pdf: r.pdf_url,
+            })),
+            note: "Use get_document with an id to retrieve the full text.",
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("get_document", "Retrieve the full text of a specific regulatory document by its id (from search results). Returns the complete body plus metadata and official link.", { id: z.string().describe("Document id from search results e.g. 'rbi:13344' or 'sebi:101703'") }, async ({ id }) => {
+    try {
+        const doc = q.getDoc(id);
+        if (!doc)
+            return err(`No document found with id ${id}. Use search_regulations to find valid ids.`);
+        const body = doc.body && doc.body.length > 12000
+            ? doc.body.slice(0, 12000) + "\n\n[... truncated. See full document at source link ...]"
+            : doc.body;
+        return ok({
+            id: doc.id, regulator: doc.regulator, type: doc.doc_type,
+            title: doc.title, date: doc.date, department: doc.department,
+            source: doc.source_url, pdf: doc.pdf_url, body,
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("get_recent", "Get the most recent regulatory documents, optionally filtered by regulator or type. Useful for 'what changed recently' questions.", {
+    regulator: z.enum(["RBI", "SEBI"]).optional(),
+    doc_type: z.enum(["circular", "master_direction", "master_circular", "notification", "regulation"]).optional(),
+    limit: z.number().default(15).describe("Max results (1-30)"),
+}, async ({ regulator, doc_type, limit }) => {
+    try {
+        if (!q.docCount().length)
+            return emptyDbMsg();
+        const docs = q.recentDocs({ regulator, docType: doc_type, limit: Math.min(limit, 30) });
+        return ok({
+            results: docs.map((d) => ({ id: d.id, regulator: d.regulator, type: d.doc_type, title: d.title, date: d.date, source: d.source_url })),
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("list_master_directions", "List RBI Master Directions and SEBI Master Circulars — the consolidated, currently-in-force rules on each subject. " +
+    "Best starting point for understanding the current state of regulation on a topic.", { regulator: z.enum(["RBI", "SEBI"]).optional() }, async ({ regulator }) => {
+    try {
+        if (!q.docCount().length)
+            return emptyDbMsg();
+        const md = q.recentDocs({ regulator, docType: "master_direction", limit: 50 });
+        const mc = q.recentDocs({ regulator, docType: "master_circular", limit: 50 });
+        const all = [...md, ...mc].sort((a, b) => b.date.localeCompare(a.date));
+        return ok({
+            count: all.length,
+            documents: all.map((d) => ({ id: d.id, regulator: d.regulator, type: d.doc_type, title: d.title, date: d.date, source: d.source_url })),
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("list_by_department", "List SEBI documents from a specific department e.g. 'Investment Management', 'Market Regulation', 'Corporation Finance'.", { department: z.string().describe("Department name or partial e.g. 'Investment Management', 'Foreign Portfolio'") }, async ({ department }) => {
+    try {
+        if (!q.docCount().length)
+            return emptyDbMsg();
+        const docs = q.listByDepartment(department, 25);
+        return ok({
+            department, count: docs.length,
+            documents: docs.map((d) => ({ id: d.id, title: d.title, date: d.date, source: d.source_url })),
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("sync_latest", "Refresh the regulatory index by scraping the latest documents from RBI and SEBI. " +
+    "Run this to pull in newly published circulars, master circulars, and regulations. Incremental — only fetches documents not already indexed. " +
+    "Note: takes 2-5 minutes as it politely scrapes the regulators' sites.", {
+    months_back: z.number().default(2).describe("How many months of RBI history to check (default 2 for incremental refresh)"),
+    sebi_pages: z.number().default(3).describe("How many SEBI listing pages to check per section (default 3 = ~75 recent docs per section)"),
+}, async ({ months_back, sebi_pages }) => {
+    try {
+        const log = [];
+        const rbiCount = await syncRbi(months_back, (m) => log.push(m));
+        const sebiCirc = await syncSebi(7, sebi_pages, (m) => log.push(m));
+        const sebiMaster = await syncSebi(6, sebi_pages, (m) => log.push(m));
+        const sebiReg = await syncSebi(3, sebi_pages, (m) => log.push(m));
+        const sebiCount = sebiCirc + sebiMaster + sebiReg;
+        q.setSyncMeta("last_sync", new Date().toISOString());
+        return ok({
+            message: "Sync complete.",
+            newRbiDocs: rbiCount, newSebiDocs: sebiCount,
+            log,
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("sync_status", "Show how many documents are indexed, broken down by regulator and type, and when the index was last synced.", {}, async () => {
+    try {
+        const counts = q.docCount();
+        const lastSync = q.getSyncMeta("last_sync");
+        const total = counts.reduce((s, c) => s + c.n, 0);
+        return ok({ totalDocuments: total, breakdown: counts, lastSync: lastSync || "never" });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+server.tool("search_by_topic", "Topic-focused search that returns BOTH the consolidated master rules AND recent circulars on a subject, " +
+    "so you get the current baseline plus any recent changes. Best tool for 'give me everything on X' questions.", { topic: z.string().describe("Regulatory topic e.g. 'digital lending', 'NBFC capital adequacy', 'FPI', 'algo trading', 'KYC'") }, async ({ topic }) => {
+    try {
+        if (!q.docCount().length)
+            return emptyDbMsg();
+        const masters = q.searchDocs({ query: topic, docType: "master_direction", limit: 3 })
+            .concat(q.searchDocs({ query: topic, docType: "master_circular", limit: 3 }));
+        const recent = q.searchDocs({ query: topic, limit: 10 });
+        return ok({
+            topic,
+            consolidatedRules: masters.map((r) => ({ id: r.id, regulator: r.regulator, title: r.title, date: r.date, source: r.source_url })),
+            relatedDocuments: recent.map((r) => ({ id: r.id, regulator: r.regulator, type: r.doc_type, title: r.title, date: r.date, snippet: r.snippet, source: r.source_url })),
+            guidance: "Start with consolidatedRules for the current baseline, then check relatedDocuments for recent amendments. Use get_document for full text.",
+            disclaimer: DISCLAIMER,
+        });
+    }
+    catch (e) {
+        return err(e instanceof Error ? e.message : String(e));
+    }
+});
+const transport = new StdioServerTransport();
+await server.connect(transport);
+console.error("india-reg-mcp running on stdio");

package/dist/scrapers/pdf.js

@@ -0,0 +1,19 @@
+import { politeFetch } from "../util/http.js";
+import { PDFParse } from "pdf-parse";
+export async function extractPdfText(pdfUrl) {
+    try {
+        const res = await politeFetch(pdfUrl);
+        const buf = Buffer.from(await res.arrayBuffer());
+        const parser = new PDFParse({ data: buf });
+        const result = await parser.getText();
+        return cleanText(result.text);
+    }
+    catch (e) {
+        const msg = e instanceof Error ? e.message : String(e);
+        console.error(`PDF extract failed for ${pdfUrl}: ${msg}`);
+        return "";
+    }
+}
+function cleanText(t) {
+    return t.replace(/\r/g, "").replace(/\n{3,}/g, "\n\n").replace(/[ \t]{2,}/g, " ").trim();
+}

package/dist/scrapers/rbi.js

@@ -0,0 +1,168 @@
+import * as cheerio from "cheerio";
+import TurndownService from "turndown";
+import pLimit from "p-limit";
+import { politeFetch, sleep } from "../util/http.js";
+import { extractPdfText } from "./pdf.js";
+import { upsertMany, docExists } from "../db/queries.js";
+const td = new TurndownService();
+const limit = pLimit(2);
+const RBI_BASE = "https://rbi.org.in/Scripts/NotificationUser.aspx";
+async function fetchViewstateTokens() {
+    const res = await politeFetch(RBI_BASE);
+    const html = await res.text();
+    return {
+        vs: html.match(/id="__VIEWSTATE"\s+value="([^"]+)"/)?.[1] ?? "",
+        vsg: html.match(/id="__VIEWSTATEGENERATOR"\s+value="([^"]+)"/)?.[1] ?? "",
+        ev: html.match(/id="__EVENTVALIDATION"\s+value="([^"]+)"/)?.[1] ?? "",
+    };
+}
+// Scrape one month via POST (month: 1-12, or 0 = all)
+export async function scrapeRbiMonth(year, month, tokens) {
+    const t = tokens ?? await fetchViewstateTokens();
+    const body = new URLSearchParams({
+        __VIEWSTATE: t.vs,
+        __VIEWSTATEGENERATOR: t.vsg,
+        __EVENTVALIDATION: t.ev,
+        hdnYear: String(year),
+        hdnMonth: String(month),
+        "UsrFontCntr$btn": "",
+    });
+    let res;
+    for (let attempt = 0; attempt <= 2; attempt++) {
+        try {
+            res = await fetch(RBI_BASE, {
+                method: "POST",
+                headers: {
+                    "User-Agent": "india-reg-mcp/1.0 (open-source regulatory indexer)",
+                    "Content-Type": "application/x-www-form-urlencoded",
+                    "Referer": RBI_BASE,
+                    "Accept": "text/html,*/*",
+                },
+                body: body.toString(),
+                signal: AbortSignal.timeout(30_000),
+            });
+            if (res.ok)
+                break;
+            if (res.status === 429 || res.status >= 500) {
+                await sleep(2000 * (attempt + 1));
+                continue;
+            }
+            throw new Error(`RBI POST failed: HTTP ${res.status}`);
+        }
+        catch (e) {
+            if (attempt === 2)
+                throw e;
+            await sleep(2000 * (attempt + 1));
+        }
+    }
+    if (!res?.ok)
+        throw new Error(`RBI POST failed after retries`);
+    const html = await res.text();
+    if (html.includes("No Notification Found"))
+        return [];
+    const $ = cheerio.load(html);
+    const items = [];
+    let currentDate = "";
+    $("table tr").each((_, tr) => {
+        const $tr = $(tr);
+        const text = $tr.text().trim();
+        const dateMatch = text.match(/^([A-Z][a-z]{2}\s+\d{1,2},\s+\d{4})$/);
+        if (dateMatch) {
+            currentDate = toISO(dateMatch[1]);
+            return;
+        }
+        const titleLink = $tr.find('a[href*="NotificationUser.aspx?Id="]').first();
+        if (titleLink.length) {
+            const href = titleLink.attr("href") || "";
+            const idMatch = href.match(/Id=(\d+)/);
+            if (!idMatch)
+                return;
+            if (!currentDate)
+                return; // skip rows before the first date header
+            const pdfLink = $tr.find('a[href*=".PDF"], a[href*=".pdf"]').first();
+            items.push({
+                id: `rbi:${idMatch[1]}`,
+                title: titleLink.text().trim(),
+                date: currentDate,
+                htmlUrl: absolute(href, "https://rbi.org.in/Scripts/"),
+                pdfUrl: pdfLink.length ? (pdfLink.attr("href") || null) : null,
+            });
+        }
+    });
+    return items;
+}
+async function fetchRbiBody(item) {
+    try {
+        const res = await politeFetch(item.htmlUrl);
+        const $ = cheerio.load(await res.text());
+        // #pnlDetails is the main content container on RBI ASP.NET doc pages
+        const main = $("#pnlDetails, #example-min, table.tablebg").first();
+        const bodyHtml = main.length ? main.html() : $("body").html();
+        let markdown = bodyHtml ? td.turndown(bodyHtml) : "";
+        if (markdown.length < 200 && item.pdfUrl)
+            markdown = await extractPdfText(item.pdfUrl);
+        return markdown;
+    }
+    catch {
+        return item.pdfUrl ? await extractPdfText(item.pdfUrl) : "";
+    }
+}
+export async function syncRbi(monthsBack, onProgress) {
+    const tokens = await fetchViewstateTokens();
+    const now = new Date();
+    let total = 0;
+    for (let i = 0; i < monthsBack; i++) {
+        const d = new Date(now.getFullYear(), now.getMonth() - i, 1);
+        const year = d.getFullYear();
+        const month = d.getMonth() + 1; // 1-indexed to match GetYearMonth JS
+        let items;
+        try {
+            items = await scrapeRbiMonth(year, month, tokens);
+        }
+        catch (e) {
+            const msg = e instanceof Error ? e.message : String(e);
+            onProgress?.(`RBI ${year}-${month}: fetch failed (${msg}), skipping`);
+            await sleep(3000);
+            continue;
+        }
+        onProgress?.(`RBI ${year}-${month}: ${items.length} docs found`);
+        const newItems = items.filter((it) => !docExists(it.id));
+        const rows = await Promise.all(newItems.map((it) => limit(async () => {
+            const body = await fetchRbiBody(it);
+            await sleep(300);
+            return {
+                id: it.id, regulator: "RBI",
+                doc_type: classifyRbi(it.title),
+                title: it.title, date: it.date, department: null,
+                source_url: it.htmlUrl, pdf_url: it.pdfUrl, body,
+                indexed_at: new Date().toISOString(),
+            };
+        })));
+        if (rows.length)
+            upsertMany(rows);
+        total += rows.length;
+        await sleep(500);
+    }
+    return total;
+}
+function classifyRbi(title) {
+    const t = title.toLowerCase();
+    if (t.includes("master direction"))
+        return "master_direction";
+    if (t.includes("master circular"))
+        return "master_circular";
+    if (t.includes("regulations"))
+        return "regulation";
+    if (t.includes("circular"))
+        return "circular";
+    return "notification";
+}
+function toISO(s) {
+    const d = new Date(s);
+    return isNaN(d.getTime()) ? "" : d.toISOString().split("T")[0];
+}
+function absolute(href, base) {
+    if (href.startsWith("http"))
+        return href;
+    return new URL(href, base).toString();
+}

package/dist/scrapers/repair-bodies.js

@@ -0,0 +1,92 @@
+/**
+ * Backfill body text for docs that were indexed without body content.
+ * Run with: npx tsx src/scrapers/repair-bodies.ts
+ */
+import { initSchema } from "../db/schema.js";
+import { db } from "../db/schema.js";
+import { sleep } from "../util/http.js";
+import pLimit from "p-limit";
+// Import body fetchers directly
+import * as cheerio from "cheerio";
+import TurndownService from "turndown";
+import { politeFetch } from "../util/http.js";
+import { PDFParse } from "pdf-parse";
+initSchema();
+const td = new TurndownService();
+const limit = pLimit(2);
+const log = (m) => console.error(`[repair] ${m}`);
+async function fetchRbiBody(sourceUrl) {
+    try {
+        const res = await politeFetch(sourceUrl);
+        const $ = cheerio.load(await res.text());
+        const main = $("#pnlDetails, #example-min, table.tablebg").first();
+        const bodyHtml = main.length ? main.html() : $("body").html();
+        return bodyHtml ? td.turndown(bodyHtml) : "";
+    }
+    catch {
+        return "";
+    }
+}
+async function fetchSebiBody(sourceUrl) {
+    try {
+        const res = await politeFetch(sourceUrl);
+        const $ = cheerio.load(await res.text());
+        const iframeSrc = $("iframe[src*='sebi_data'], iframe[src*='?file=']").first().attr("src") || "";
+        const pdfUrlMatch = iframeSrc.match(/[?&]file=((?:https?:\/\/|\/)[^'"&\s]+\.pdf)/i);
+        const rawPdfPath = pdfUrlMatch ? pdfUrlMatch[1] : null;
+        const pdfUrl = rawPdfPath
+            ? rawPdfPath.startsWith("/") ? `https://www.sebi.gov.in${rawPdfPath}` : rawPdfPath
+            : null;
+        if (pdfUrl) {
+            const buf = Buffer.from(await (await politeFetch(pdfUrl)).arrayBuffer());
+            const parser = new PDFParse({ data: buf });
+            const result = await parser.getText();
+            const text = result.text.replace(/\r/g, "").replace(/\n{3,}/g, "\n\n").trim();
+            return { body: text, pdfUrl };
+        }
+        return { body: "", pdfUrl: null };
+    }
+    catch {
+        return { body: "", pdfUrl: null };
+    }
+}
+const updateStmt = db.prepare("UPDATE documents SET body=@body, pdf_url=@pdf_url, indexed_at=@indexed_at WHERE id=@id");
+const updateBodyOnly = db.prepare("UPDATE documents SET body=@body, indexed_at=@indexed_at WHERE id=@id");
+const docs = db.prepare("SELECT id, regulator, source_url FROM documents WHERE body IS NULL OR body = '' OR (regulator='SEBI' AND LENGTH(body) < 2000) ORDER BY date DESC").all();
+log(`Found ${docs.length} docs needing body repair`);
+let done = 0;
+let failed = 0;
+await Promise.all(docs.map((doc) => limit(async () => {
+    try {
+        let body = "";
+        let pdfUrl = null;
+        if (doc.regulator === "RBI") {
+            body = await fetchRbiBody(doc.source_url);
+        }
+        else {
+            const result = await fetchSebiBody(doc.source_url);
+            body = result.body;
+            pdfUrl = result.pdfUrl;
+        }
+        if (doc.regulator === "RBI") {
+            updateBodyOnly.run({ id: doc.id, body, indexed_at: new Date().toISOString() });
+        }
+        else {
+            updateStmt.run({ id: doc.id, body, pdf_url: pdfUrl, indexed_at: new Date().toISOString() });
+        }
+        done++;
+        if (done % 20 === 0)
+            log(`Progress: ${done}/${docs.length} done`);
+    }
+    catch (e) {
+        const msg = e instanceof Error ? e.message : String(e);
+        console.error(`Failed ${doc.id}: ${msg}`);
+        failed++;
+    }
+    await sleep(300);
+})));
+log(`Done. ${done} updated, ${failed} failed.`);
+// Show updated stats
+const stats = db.prepare("SELECT regulator, SUM(CASE WHEN body IS NULL OR body='' THEN 1 ELSE 0 END) as no_body, COUNT(*) as total FROM documents GROUP BY regulator").all();
+stats.forEach(s => log(`${s.regulator}: ${s.total - s.no_body}/${s.total} have body`));
+process.exit(0);

package/dist/scrapers/run-sync.js

@@ -0,0 +1,31 @@
+import { initSchema } from "../db/schema.js";
+import { syncRbi } from "./rbi.js";
+import { syncSebi } from "./sebi.js";
+import { setSyncMeta } from "../db/queries.js";
+async function main() {
+    initSchema();
+    const log = (m) => console.error(`[sync] ${m}`);
+    const args = process.argv.slice(2);
+    const quick = args.includes("--quick"); // quick mode: 6mo RBI + 5 pages SEBI
+    if (quick) {
+        log("Quick sync mode (6 months RBI, 5 pages SEBI each)...");
+        const rbiCount = await syncRbi(6, log);
+        log(`RBI: ${rbiCount} new documents`);
+        const sebiCirc = await syncSebi(7, 5, log);
+        log(`SEBI circulars: ${sebiCirc} new documents`);
+    }
+    else {
+        log("Starting RBI sync (last 36 months)...");
+        const rbiCount = await syncRbi(36, log);
+        log(`RBI: ${rbiCount} new documents`);
+        log("Starting SEBI sync...");
+        const sebiMaster = await syncSebi(6, 5, log);
+        const sebiCirc = await syncSebi(7, 40, log);
+        const sebiReg = await syncSebi(3, 10, log);
+        log(`SEBI: ${sebiMaster + sebiCirc + sebiReg} new documents`);
+    }
+    setSyncMeta("last_sync", new Date().toISOString());
+    log("Sync complete.");
+    process.exit(0);
+}
+main().catch((e) => { console.error(e); process.exit(1); });

package/dist/scrapers/sebi.js

@@ -0,0 +1,146 @@
+import * as cheerio from "cheerio";
+import TurndownService from "turndown";
+import pLimit from "p-limit";
+import { politeFetch, sleep } from "../util/http.js";
+import { upsertMany, docExists } from "../db/queries.js";
+const td = new TurndownService();
+const limit = pLimit(2);
+const UA = "india-reg-mcp/1.0 (open-source regulatory indexer)";
+// ssid: 7=circulars, 6=master circulars, 3=regulations
+const SEBI_LIST_BASE = "https://www.sebi.gov.in/sebiweb/home/HomeAction.do?doListing=yes&sid=1&ssid=";
+const SEBI_AJAX = "https://www.sebi.gov.in/sebiweb/ajax/home/getnewslistinfo.jsp";
+function parseListItems($) {
+    const items = [];
+    $("table tr").each((_, tr) => {
+        const $tr = $(tr);
+        const link = $tr.find('a[href*="/legal/"]').first();
+        if (!link.length)
+            return;
+        const href = link.attr("href") || "";
+        const idMatch = href.match(/_(\d+)\.html/);
+        if (!idMatch)
+            return;
+        const dateCell = $tr.find("td").first().text().trim();
+        items.push({
+            id: `sebi:${idMatch[1]}`,
+            title: link.text().trim(),
+            date: toISO(dateCell),
+            url: absolute(href, "https://www.sebi.gov.in/"),
+        });
+    });
+    return items;
+}
+// Pages 1+: AJAX POST to getnewslistinfo.jsp (page 0 handled in syncSebi)
+async function getSebiPage(ssid, pageIndex, jsessionid) {
+    // Pages 1+: AJAX POST to getnewslistinfo.jsp
+    const body = new URLSearchParams({
+        nextValue: "1",
+        next: "n",
+        search: "", fromDate: "", toDate: "", fromYear: "", toYear: "",
+        deptId: "",
+        sid: "1", ssid: String(ssid), smid: "0", ssidhidden: String(ssid),
+        intmid: "-1",
+        sText: "Legal", ssText: ssid === 7 ? "Circulars" : ssid === 6 ? "Master Circulars" : "Regulations",
+        smText: "",
+        doDirect: String(pageIndex),
+    });
+    const res = await fetch(SEBI_AJAX, {
+        method: "POST",
+        headers: {
+            "User-Agent": UA,
+            "Content-Type": "application/x-www-form-urlencoded",
+            "Cookie": `JSESSIONID=${jsessionid}`,
+            "Referer": `${SEBI_LIST_BASE}${ssid}&smid=0&nextValue=0`,
+            "Accept": "*/*",
+            "X-Requested-With": "XMLHttpRequest",
+        },
+        body: body.toString(),
+        signal: AbortSignal.timeout(30_000),
+    });
+    if (!res.ok)
+        throw new Error(`SEBI AJAX failed: HTTP ${res.status}`);
+    return parseListItems(cheerio.load(await res.text()));
+}
+async function fetchSebiBody(item) {
+    try {
+        const res = await politeFetch(item.url);
+        const $ = cheerio.load(await res.text());
+        // SEBI pages embed content as PDF in an iframe — src may have absolute or relative PDF path
+        // e.g. ?file=https://www.sebi.gov.in/sebi_data/... or ?file=/sebi_data/...
+        const iframeSrc = $("iframe[src*='sebi_data'], iframe[src*='?file=']").first().attr("src") || "";
+        const pdfUrlMatch = iframeSrc.match(/[?&]file=((?:https?:\/\/|\/)[^'"&\s]+\.pdf)/i);
+        const rawPdfPath = pdfUrlMatch ? pdfUrlMatch[1] : null;
+        const pdfUrl = rawPdfPath
+            ? rawPdfPath.startsWith("/") ? `https://www.sebi.gov.in${rawPdfPath}` : rawPdfPath
+            : null;
+        if (pdfUrl) {
+            const { extractPdfText } = await import("./pdf.js");
+            const body = await extractPdfText(pdfUrl);
+            return { body, pdfUrl };
+        }
+        // Fallback: extract any visible text from the page
+        const main = $(".main_section, .news-detail-slider, #member-wrapper").first();
+        const bodyHtml = main.length ? main.html() : $("body").html();
+        return { body: bodyHtml ? td.turndown(bodyHtml) : "", pdfUrl: null };
+    }
+    catch {
+        return { body: "", pdfUrl: null };
+    }
+}
+const SSID_DOC_TYPE = { 6: "master_circular", 3: "regulation" };
+const SSID_DEPARTMENT = { 6: "Master Circulars", 3: "Regulations", 7: "Circulars" };
+export async function syncSebi(ssid, maxPages, onProgress) {
+    // Page 0: single GET that both establishes the session and returns first page listings
+    const url = `${SEBI_LIST_BASE}${ssid}&smid=0&nextValue=0`;
+    const page0Res = await fetch(url, {
+        headers: { "User-Agent": UA, "Accept": "text/html,*/*" },
+        signal: AbortSignal.timeout(30_000),
+    });
+    if (!page0Res.ok)
+        throw new Error(`SEBI GET failed: HTTP ${page0Res.status}`);
+    const cookie = page0Res.headers.get("set-cookie") || "";
+    const jsessionid = cookie.match(/JSESSIONID=([^;]+)/)?.[1] ?? "";
+    if (!jsessionid)
+        throw new Error("SEBI: failed to obtain JSESSIONID — session cookie absent from page-0 response");
+    const doc_type = SSID_DOC_TYPE[ssid] ?? "circular";
+    const department = SSID_DEPARTMENT[ssid] ?? "Circulars";
+    let total = 0;
+    const processPage = async (items) => {
+        if (!items.length)
+            return false;
+        const newItems = items.filter((it) => !docExists(it.id));
+        const rows = await Promise.all(newItems.map((it) => limit(async () => {
+            const { body, pdfUrl } = await fetchSebiBody(it);
+            await sleep(300);
+            return {
+                id: it.id, regulator: "SEBI",
+                doc_type, title: it.title, date: it.date, department,
+                source_url: it.url, pdf_url: pdfUrl, body,
+                indexed_at: new Date().toISOString(),
+            };
+        })));
+        if (rows.length)
+            upsertMany(rows);
+        total += rows.length;
+        await sleep(500);
+        return true;
+    };
+    const page0Items = parseListItems(cheerio.load(await page0Res.text()));
+    onProgress?.(`SEBI ssid=${ssid} page 0: ${page0Items.length} docs`);
+    await processPage(page0Items);
+    for (let page = 1; page < maxPages; page++) {
+        const items = await getSebiPage(ssid, page, jsessionid);
+        if (!items.length)
+            break;
+        onProgress?.(`SEBI ssid=${ssid} page ${page}: ${items.length} docs`);
+        await processPage(items);
+    }
+    return total;
+}
+function toISO(s) {
+    const d = new Date(s.replace(/(\w{3})\s+(\d{1,2}),\s+(\d{4})/, "$1 $2 $3"));
+    return isNaN(d.getTime()) ? "" : d.toISOString().split("T")[0];
+}
+function absolute(href, base) {
+    return href.startsWith("http") ? href : new URL(href, base).toString();
+}

package/dist/util/format.js

@@ -0,0 +1,10 @@
+export const DISCLAIMER = "Source: official RBI/SEBI publications. This is primary-source retrieval, not legal advice. Verify against the linked official document.";
+export function ok(data) {
+    return { content: [{ type: "text", text: JSON.stringify(data, null, 2) }] };
+}
+export function err(m) {
+    return { content: [{ type: "text", text: `Error: ${m}` }] };
+}
+export function emptyDbMsg() {
+    return ok({ message: "The regulatory index is empty. Run 'npm run sync' first, or call the sync_latest tool to populate it.", disclaimer: DISCLAIMER });
+}

package/dist/util/http.js

@@ -0,0 +1,25 @@
+const UA = "india-reg-mcp/1.0 (open-source regulatory indexer; +https://github.com/yourusername/india-reg-mcp)";
+export async function politeFetch(url, retries = 2) {
+    for (let attempt = 0; attempt <= retries; attempt++) {
+        try {
+            const res = await fetch(url, {
+                headers: { "User-Agent": UA, "Accept": "text/html,application/pdf,*/*" },
+                signal: AbortSignal.timeout(30_000),
+            });
+            if (res.ok)
+                return res;
+            if (res.status === 429 || res.status >= 500) {
+                await sleep(1000 * (attempt + 1));
+                continue;
+            }
+            throw new Error(`HTTP ${res.status} for ${url}`);
+        }
+        catch (e) {
+            if (attempt === retries)
+                throw e;
+            await sleep(1000 * (attempt + 1));
+        }
+    }
+    throw new Error(`Failed after ${retries} retries: ${url}`);
+}
+export function sleep(ms) { return new Promise((r) => setTimeout(r, ms)); }

package/package.json

@@ -0,0 +1,61 @@
+{
+  "name": "india-reg-mcp",
+  "version": "1.0.0",
+  "description": "MCP server for Indian financial regulations — RBI & SEBI circulars, master directions, notifications. Searchable, cited, no API keys.",
+  "type": "module",
+  "main": "dist/index.js",
+  "bin": {
+    "india-reg-mcp": "./dist/index.js"
+  },
+  "files": [
+    "dist/**/*",
+    "README.md"
+  ],
+  "keywords": [
+    "mcp",
+    "model-context-protocol",
+    "india",
+    "rbi",
+    "sebi",
+    "regulations",
+    "circulars",
+    "compliance",
+    "fintech",
+    "claude"
+  ],
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/Akhilgovind02/india-regulatory-mcp.git"
+  },
+  "homepage": "https://github.com/Akhilgovind02/india-regulatory-mcp",
+  "engines": {
+    "node": ">=18.0.0"
+  },
+  "scripts": {
+    "build": "tsc",
+    "postbuild": "node scripts/add-shebang.mjs",
+    "dev": "tsx src/index.ts",
+    "sync": "tsx src/scrapers/run-sync.ts",
+    "inspect": "npx @modelcontextprotocol/inspector tsx src/index.ts",
+    "start": "node dist/index.js",
+    "prepublishOnly": "npm run build"
+  },
+  "dependencies": {
+    "@modelcontextprotocol/sdk": "1.29.0",
+    "better-sqlite3": "12.10.0",
+    "cheerio": "1.2.0",
+    "p-limit": "7.3.0",
+    "pdf-parse": "2.4.5",
+    "turndown": "7.2.4",
+    "zod": "4.4.3"
+  },
+  "devDependencies": {
+    "@modelcontextprotocol/inspector": "0.22.0",
+    "@types/better-sqlite3": "^7.6.13",
+    "@types/node": "25.9.3",
+    "@types/turndown": "^5.0.6",
+    "tsx": "4.22.4",
+    "typescript": "6.0.3"
+  }
+}