AGENT SKILL

detecting-modbus-command-injection-attacks

Detect command injection against Modbus TCP/RTU in ICS/SCADA environments by monitoring unauthorized writes, anomalous function codes, malformed frames, and deviations from communication baselines using ICS-aware IDS and deep packet inspection. Use when deploying IDS for Modbus OT networks, investigating unauthorized PLC register/coil changes, or responding to FrostyGoop-style Modbus attacks.

License 许可
Apache-2.0
最近更新
2026-08-25
Stars
★ 0
Forks
0

01 前言 Front Matter

name
detecting-modbus-command-injection-attacks
tags
[ "ot-security", "ics", "modbus", "command-injection", "protocol-analysis", "ids", "scada", "threat-detection" ]
author
mahipal
domain
cybersecurity
license
Apache-2.0
version
1.0
nist_csf
[ "PR.IR-01", "DE.CM-01", "ID.AM-05", "GV.OC-02" ]
subdomain
ot-ics-security
description
Detect command injection against Modbus TCP/RTU in ICS/SCADA environments by monitoring unauthorized writes, anomalous function codes, malformed frames, and deviations from communication baselines using ICS-aware IDS and deep packet inspection. Use when deploying IDS for Modbus OT networks, investigating unauthorized PLC register/coil changes, or responding to FrostyGoop-style Modbus attacks.
mitre_attack
[ "T1078", "T1190", "T1059", "T1055", "T0816" ]

02 资源文件 3

03 SKILL.md 正文

Detecting Modbus Command Injection Attacks

When to Use

  • When deploying intrusion detection for environments using Modbus TCP (port 502) or Modbus RTU
  • When investigating suspected unauthorized modifications to PLC registers or coils
  • When building detection analytics for OT SOC monitoring Modbus-heavy environments
  • When responding to FrostyGoop-style attacks that leverage Modbus TCP for operational impact
  • When performing baseline validation after a suspected compromise of a Modbus master

Do not use for detecting attacks on non-Modbus protocols (see detecting-dnp3-protocol-anomalies for DNP3), for general IT network intrusion detection, or for Modbus device configuration (see performing-ot-vulnerability-scanning-safely).

Prerequisites

  • Network SPAN/TAP on the segment carrying Modbus TCP traffic (typically port 502)
  • Baseline of normal Modbus communication patterns (masters, slaves, function codes, register ranges, polling intervals)
  • Suricata, Zeek, or commercial OT IDS deployed with Modbus protocol parsers enabled
  • Understanding of Modbus function codes used in the environment (read vs write operations)
  • Access to PLC programming documentation to validate expected register ranges

Workflow

Step 1: Build Modbus Communication Baseline

Capture and analyze normal Modbus traffic to establish what constitutes legitimate communication patterns.

#!/usr/bin/env python3
"""Modbus Command Injection Detector.

Monitors Modbus TCP traffic for unauthorized write operations, anomalous
function codes, and deviations from established communication baselines.
Detects attacks like FrostyGoop that use Modbus TCP for operational impact.
"""

import json
import struct
import sys
import time
from collections import defaultdict
from datetime import datetime
from typing import Dict, List, Optional, Set, Tuple

try:
    from scapy.all import sniff, IP, TCP
except ImportError:
    print("Install scapy: pip install scapy")
    sys.exit(1)


# Modbus function code definitions
MODBUS_READ_FUNCTIONS = {1, 2, 3, 4}
MODBUS_WRITE_FUNCTIONS = {5, 6, 15, 16}
MODBUS_DIAGNOSTIC_FUNCTIONS = {8, 17, 43}

MODBUS_FUNC_NAMES = {
    1: "Read Coils", 2: "Read Discrete Inputs",
    3: "Read Holding Registers", 4: "Read Input Registers",
    5: "Write Single Coil", 6: "Write Single Register",
    8: "Diagnostics", 15: "Write Multiple Coils",
    16: "Write Multiple Registers", 17: "Report Slave ID",
    22: "Mask Write Register", 23: "Read/Write Multiple Registers",
    43: "Encapsulated Interface Transport",
}


class ModbusAlert:
    """Represents a detected Modbus anomaly."""

    def __init__(self, severity: str, alert_type: str, src_ip: str,
                 dst_ip: str, unit_id: int, func_code: int,
                 description: str, mitre_technique: str = ""):
        self.timestamp = datetime.now().isoformat()
        self.severity = severity
        self.alert_type = alert_type
        self.src_ip = src_ip
        self.dst_ip = dst_ip
        self.unit_id = unit_id
        self.func_code = func_code
        self.func_name = MODBUS_FUNC_NAMES.get(func_code, f"Unknown FC {func_code}")
        self.description = description
        self.mitre_technique = mitre_technique

    def __str__(self):
        return (
            f"[{self.severity}] {self.alert_type} | {self.src_ip} -> {self.dst_ip} "
            f"| Unit {self.unit_id} | {self.func_name} | {self.description}"
        )


class ModbusInjectionDetector:
    """Detects Modbus command injection attacks."""

    def __init__(self, baseline_file: Optional[str] = None):
        self.alerts: List[ModbusAlert] = []
        self.packet_count = 0
        self.modbus_count = 0

        # Baseline data
        self.authorized_masters: Set[str] = set()
        self.authorized_pairs: Set[Tuple[str, str]] = set()
        self.allowed_write_sources: Set[str] = set()
        self.allowed_function_codes: Dict[str, Set[int]] = defaultdict(set)
        self.allowed_register_ranges: Dict[str, List[Tuple[int, int]]] = defaultdict(list)
        self.polling_intervals: Dict[str, float] = {}
        self.last_seen: Dict[str, float] = {}

        # Counters for rate detection
        self.write_counts: Dict[str, List[float]] = defaultdict(list)

        if baseline_file:
            self.load_baseline(baseline_file)

    def load_baseline(self, filepath: str):
        """Load established Modbus communication baseline."""
        with open(filepath, "r") as f:
            baseline = json.load(f)

        for session_key, data in baseline.get("modbus_baselines", {}).items():
            src, dst = session_key.split("->")
            self.authorized_pairs.add((src.strip(), dst.strip()))
            self.authorized_masters.add(src.strip())

            fc_set = set(data.get("allowed_function_codes", []))
            self.allowed_function_codes[session_key] = fc_set

            if fc_set & MODBUS_WRITE_FUNCTIONS:
                self.allowed_write_sources.add(src.strip())

            for reg_range in data.get("register_ranges", []):
                self.allowed_register_ranges[session_key].append(
                    (reg_range["start"], reg_range["end"])
                )

            if data.get("polling_interval_avg_sec"):
                self.polling_intervals[session_key] = data["polling_interval_avg_sec"]

        print(f"[*] Baseline loaded: {len(self.authorized_pairs)} authorized pairs, "
              f"{len(self.allowed_write_sources)} authorized write sources")

    def parse_modbus_mbap(self, payload: bytes) -> Optional[dict]:
        """Parse Modbus TCP MBAP header and PDU."""
        if len(payload) < 8:
            return None

        transaction_id = struct.unpack(">H", payload[0:2])[0]
        protocol_id = struct.unpack(">H", payload[2:4])[0]
        length = struct.unpack(">H", payload[4:6])[0]
        unit_id = payload[6]
        func_code = payload[7]

        if protocol_id != 0:  # Not Modbus
            return None

        result = {
            "transaction_id": transaction_id,
            "protocol_id": protocol_id,
            "length": length,
            "unit_id": unit_id,
            "func_code": func_code,
        }

        # Parse register address and count for read/write operations
        if len(payload) >= 12 and func_code in (1, 2, 3, 4, 5, 6, 15, 16):
            result["start_address"] = struct.unpack(">H", payload[8:10])[0]
            result["quantity"] = struct.unpack(">H", payload[10:12])[0]

        return result

    def analyze_packet(self, pkt):
        """Analyze a network packet for Modbus command injection."""
        self.packet_count += 1

        if not pkt.haslayer(IP) or not pkt.haslayer(TCP):
            return

        tcp = pkt[TCP]
        if tcp.dport != 502 and tcp.sport != 502:
            return

        payload = bytes(tcp.payload)
        if not payload:
            return

        modbus = self.parse_modbus_mbap(payload)
        if not modbus:
            return

        self.modbus_count += 1
        src_ip = pkt[IP].src
        dst_ip = pkt[IP].dst
        session_key = f"{src_ip}->{dst_ip}"
        now = time.time()

        # Detection Rule 1: Unauthorized Modbus master
        if self.authorized_masters and src_ip not in self.authorized_masters:
            if tcp.dport == 502:
                self.alerts.append(ModbusAlert(
                    severity="CRITICAL",
                    alert_type="UNAUTHORIZED_MASTER",
                    src_ip=src_ip, dst_ip=dst_ip,
                    unit_id=modbus["unit_id"],
                    func_code=modbus["func_code"],
                    description=f"Unauthorized device {src_ip} sending Modbus commands to {dst_ip}",
                    mitre_technique="T0843 - Program Download",
                ))

        # Detection Rule 2: Unauthorized write operation
        if modbus["func_code"] in MODBUS_WRITE_FUNCTIONS:
            if self.allowed_write_sources and src_ip not in self.allowed_write_sources:
                self.alerts.append(ModbusAlert(
                    severity="CRITICAL",
                    alert_type="UNAUTHORIZED_WRITE",
                    src_ip=src_ip, dst_ip=dst_ip,
                    unit_id=modbus["unit_id"],
                    func_code=modbus["func_code"],
                    description=f"Write command from non-authorized source {src_ip}",
                    mitre_technique="T0855 - Unauthorized Command Message",
                ))

            # Track write frequency for rate anomaly detection
            self.write_counts[src_ip].append(now)
            recent_writes = [t for t in self.write_counts[src_ip] if now - t < 60]
            self.write_counts[src_ip] = recent_writes
            if len(recent_writes) > 20:
                self.alerts.append(ModbusAlert(
                    severity="HIGH",
                    alert_type="WRITE_FLOOD",
                    src_ip=src_ip, dst_ip=dst_ip,
                    unit_id=modbus["unit_id"],
                    func_code=modbus["func_code"],
                    description=f"Excessive write rate: {len(recent_writes)} writes in 60s from {src_ip}",
                    mitre_technique="T0836 - Modify Parameter",
                ))

        # Detection Rule 3: Anomalous function code
        if session_key in self.allowed_function_codes:
            if modbus["func_code"] not in self.allowed_function_codes[session_key]:
                self.alerts.append(ModbusAlert(
                    severity="HIGH",
                    alert_type="ANOMALOUS_FUNCTION_CODE",
                    src_ip=src_ip, dst_ip=dst_ip,
                    unit_id=modbus["unit_id"],
                    func_code=modbus["func_code"],
                    description=(
                        f"Function code {modbus['func_code']} ({MODBUS_FUNC_NAMES.get(modbus['func_code'], 'Unknown')}) "
                        f"not in baseline for {session_key}"
                    ),
                    mitre_technique="T0855 - Unauthorized Command Message",
                ))

        # Detection Rule 4: Broadcast write (unit ID 0)
        if modbus["unit_id"] == 0 and modbus["func_code"] in MODBUS_WRITE_FUNCTIONS:
            self.alerts.append(ModbusAlert(
                severity="CRITICAL",
                alert_type="BROADCAST_WRITE",
                src_ip=src_ip, dst_ip=dst_ip,
                unit_id=0,
                func_code=modbus["func_code"],
                description="Broadcast write command (unit ID 0) affects ALL Modbus devices on segment",
                mitre_technique="T0855 - Unauthorized Command Message",
            ))

        # Detection Rule 5: Out-of-range register access
        if "start_address" in modbus and session_key in self.allowed_register_ranges:
            addr = modbus["start_address"]
            qty = modbus.get("quantity", 1)
            in_range = any(
                start <= addr and addr + qty <= end
                for start, end in self.allowed_register_ranges[session_key]
            )
            if not in_range:
                self.alerts.append(ModbusAlert(
                    severity="HIGH",
                    alert_type="OUT_OF_RANGE_REGISTER",
                    src_ip=src_ip, dst_ip=dst_ip,
                    unit_id=modbus["unit_id"],
                    func_code=modbus["func_code"],
                    description=f"Register access {addr}-{addr+qty} outside baseline ranges",
                    mitre_technique="T0836 - Modify Parameter",
                ))

        # Detection Rule 6: Diagnostic/restart commands
        if modbus["func_code"] in MODBUS_DIAGNOSTIC_FUNCTIONS:
            self.alerts.append(ModbusAlert(
                severity="HIGH",
                alert_type="DIAGNOSTIC_COMMAND",
                src_ip=src_ip, dst_ip=dst_ip,
                unit_id=modbus["unit_id"],
                func_code=modbus["func_code"],
                description=f"Diagnostic function code {modbus['func_code']} detected - potential DoS or reconnaissance",
                mitre_technique="T0814 - Denial of Service",
            ))

    def print_report(self):
        """Print detection report."""
        print(f"\n{'='*70}")
        print(f"MODBUS COMMAND INJECTION DETECTION REPORT")
        print(f"{'='*70}")
        print(f"Analysis Time: {datetime.now().isoformat()}")
        print(f"Total Packets Analyzed: {self.packet_count}")
        print(f"Modbus Packets: {self.modbus_count}")
        print(f"Alerts Generated: {len(self.alerts)}")

        if self.alerts:
            severity_counts = defaultdict(int)
            for alert in self.alerts:
                severity_counts[alert.severity] += 1

            print(f"\nSeverity Distribution:")
            for sev in ["CRITICAL", "HIGH", "MEDIUM", "LOW"]:
                if sev in severity_counts:
                    print(f"  {sev}: {severity_counts[sev]}")

            print(f"\nDetailed Alerts:")
            for alert in self.alerts:
                print(f"\n  [{alert.severity}] {alert.alert_type}")
                print(f"    Time: {alert.timestamp}")
                print(f"    Source: {alert.src_ip} -> {alert.dst_ip}")
                print(f"    Unit ID: {alert.unit_id}")
                print(f"    Function: {alert.func_name} (FC {alert.func_code})")
                print(f"    Detail: {alert.description}")
                if alert.mitre_technique:
                    print(f"    MITRE ATT&CK ICS: {alert.mitre_technique}")

    def start_live_monitoring(self, interface: str, duration: int = 0):
        """Start live Modbus traffic monitoring."""
        print(f"[*] Starting Modbus monitoring on {interface}...")
        print(f"[*] Press Ctrl+C to stop")
        try:
            sniff(
                iface=interface,
                filter="tcp port 502",
                prn=self.analyze_packet,
                timeout=duration if duration > 0 else None,
            )
        except KeyboardInterrupt:
            pass
        self.print_report()


if __name__ == "__main__":
    detector = ModbusInjectionDetector(
        baseline_file=sys.argv[2] if len(sys.argv) > 2 else None
    )

    if len(sys.argv) >= 2:
        if sys.argv[1].endswith(".pcap") or sys.argv[1].endswith(".pcapng"):
            from scapy.all import rdpcap
            print(f"[*] Analyzing capture file: {sys.argv[1]}")
            packets = rdpcap(sys.argv[1])
            for pkt in packets:
                detector.analyze_packet(pkt)
            detector.print_report()
        else:
            detector.start_live_monitoring(sys.argv[1])
    else:
        print("Usage:")
        print("  Live:    python modbus_detector.py <interface> [baseline.json]")
        print("  Offline: python modbus_detector.py <capture.pcap> [baseline.json]")

Step 2: Deploy Suricata Rules for Modbus Attack Detection

# Suricata IDS Rules for Modbus Command Injection Detection
# Reference: MITRE ATT&CK for ICS, FrostyGoop analysis

# Unauthorized Modbus write from non-engineering workstation
alert modbus !$MODBUS_AUTHORIZED_WRITERS any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Unauthorized write operation detected";
  modbus_func:write_single_coil;
  flow:to_server,established;
  classtype:attempted-admin;
  sid:4000001; rev:1; priority:1;
)

alert modbus !$MODBUS_AUTHORIZED_WRITERS any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Unauthorized write multiple registers";
  modbus_func:write_multiple_registers;
  flow:to_server,established;
  classtype:attempted-admin;
  sid:4000002; rev:1; priority:1;
)

# Modbus broadcast write affecting all slaves
alert modbus any any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Broadcast write command (Unit ID 0)";
  modbus_unit_id:0;
  flow:to_server,established;
  classtype:attempted-admin;
  sid:4000003; rev:1; priority:1;
)

# Excessive Modbus write rate (potential automated attack)
alert modbus any any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Excessive write rate - possible automated attack";
  modbus_func:write_multiple_registers;
  flow:to_server,established;
  threshold:type threshold, track by_src, count 20, seconds 60;
  classtype:attempted-admin;
  sid:4000004; rev:1;
)

# Modbus diagnostics/restart command
alert modbus any any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Diagnostics function code detected";
  modbus_func:diagnostics;
  flow:to_server,established;
  classtype:attempted-dos;
  sid:4000005; rev:1;
)

# FrostyGoop-pattern: write to specific register ranges used for heating control
alert modbus any any -> $OT_PLC_SUBNET 502 (
  msg:"MODBUS-INJECT Potential FrostyGoop - write to heating control registers";
  modbus_func:write_multiple_registers;
  content:"|00 10|"; offset:8; depth:2;
  flow:to_server,established;
  classtype:attempted-admin;
  sid:4000010; rev:1; priority:1;
)

Key Concepts

Term Definition
Modbus TCP Industrial protocol operating on TCP port 502, lacking authentication or encryption, making it vulnerable to command injection
Function Code Single byte in Modbus PDU specifying the operation (read coils, write registers, diagnostics); monitoring for unauthorized function codes is key to detection
MBAP Header Modbus Application Protocol header in TCP variant containing transaction ID, protocol ID, length, and unit ID
FrostyGoop First known malware using Modbus TCP for real-world operational impact, disrupted Ukrainian district heating in 2024
Unit ID Address of the target Modbus slave device; Unit ID 0 is a broadcast affecting all slaves
Register Range Specific memory addresses in the PLC; legitimate operations access known ranges; out-of-range access indicates reconnaissance or manipulation

Common Scenarios

Scenario: FrostyGoop-Style Heating Control Attack

Context: A building automation system uses Modbus TCP to control HVAC equipment. Monitoring detects unexpected write commands to heating control registers from an IP not associated with any authorized BMS controller.

Approach:

  1. Verify the source IP against the authorized Modbus master list
  2. Check if any authorized maintenance or configuration change is in progress
  3. Capture full Modbus transaction including register addresses and values being written
  4. Compare written values against safe operating ranges for the heating equipment
  5. If unauthorized, immediately block the source IP at the industrial firewall
  6. Inspect the source device for compromise indicators (malware, unauthorized remote access)
  7. Verify current setpoints on all affected controllers against known-good values
  8. Restore safe setpoints if manipulation is confirmed

Pitfalls: Modbus lacks authentication, so the source IP is the only identifier -- attackers can spoof IPs if ARP protections are not in place. Do not assume all writes are malicious; legitimate SCADA operations include writes. Always verify against the change management log before escalating.

Output Format

MODBUS INJECTION DETECTION REPORT
====================================
Analysis Period: [start] to [end]
Monitoring Point: [interface/SPAN description]

TRAFFIC SUMMARY:
  Total Modbus Packets: [count]
  Read Operations: [count]
  Write Operations: [count]
  Unauthorized Writes Detected: [count]

ALERTS:
  [CRITICAL] Unauthorized write from [IP] to PLC [IP]
    Function: Write Multiple Registers (FC 16)
    Registers: [start]-[end]
    MITRE: T0855 - Unauthorized Command Message

BASELINE DEVIATIONS:
  New Modbus masters: [list]
  Unusual function codes: [list]
  Out-of-range register access: [list]

RECOMMENDED ACTIONS:
  1. Verify source [IP] authorization status
  2. Block unauthorized sources at industrial firewall
  3. Validate PLC register values against known-good state

04 同仓库更多 Skills

vikasvardhanv
vikasvardhanv/jerry_skills

syllabus

Generates a curated supplementary reading list from any course syllabus using Consensus academic search. Grill-me intake (syllabus input format + course audience + year range) plus a grouping forcing-options checkpoint before any search runs — so the reading list matches the course's level and recency need. Parses the syllabus to extract topics and learning outcomes, searches Consensus for recent peer-reviewed papers per topic, and produces a professionally formatted .docx with clickable Consensus links, plain-language summaries calibrated to audience level, and Bloom-higher-order discussion questions tied to course learning goals. Use when the user uploads a syllabus, course outline, or curriculum document and wants supplementary readings (e.g., 'create a reading list from this syllabus', 'find recent papers for my course') — even casual mentions with a syllabus attached should trigger this skill.

★ 0
vikasvardhanv
vikasvardhanv/jerry_skills

pulse

Multi-source recency research skill that takes the pulse of any topic across Reddit, Hacker News, the open web, and optionally X/Twitter within a configurable recent window (default 30 days). Forcing intake clarifies topic specificity, angle (trend/sentiment/problems/opportunities/comparison), time window, and platform scope before searching. Returns a synthesized briefing with citations, engagement metrics, and cross-platform pattern analysis. Use when the user requests multi-source recency intelligence on a topic (e.g., 'pulse on [topic]', 'what's happening with [topic]', 'what are people saying about [topic]', 'current conversation about [topic]', 'take the pulse of [topic]', 'trending: [topic]', 'find me info on [topic]'), and for competitor research, trend discovery, tool comparisons, and audience sentiment analysis.

★ 0
vikasvardhanv
vikasvardhanv/jerry_skills

patent

Patent prior-art and landscape intelligence skill — not generic patent help. Commits to one of five sub-use-cases via forcing intake (novelty search / freedom-to-operate / competitive landscape / acquisition diligence / litigation prior-art) before any search runs. Searches Google Patents, Espacenet, USPTO, and optionally Lens.org for citation-graph signals. Output is an editable Word document (.docx) with verdict, ranked closest art (claim-text extracted), CPC-class-aware landscape, family-resolved hits, geographic coverage, FTO flags where applicable, strategy recommendations, and full audit log. Use when the user asks for patent searching or analysis (e.g., 'prior art search for [invention]', 'freedom to operate analysis for [product]'). Produces search signal, not legal advice — always recommends consulting a patent attorney before filing or licensing decisions. Trademark, copyright, and trade-secret questions are out of scope.

★ 0
vikasvardhanv
vikasvardhanv/jerry_skills

notebooklm

Browser automation skill for controlling Google's NotebookLM. Use when the user wants anything done in NotebookLM (e.g., 'open NotebookLM', 'check my [name] notebook', 'ask my notebook about X', 'add [source] to NotebookLM', 'generate a Video Overview from my notebook', 'use NotebookLM Studio'). Handles reading and querying notebooks, adding sources (URLs, text, files, YouTube links, synthesized content), generating Studio outputs (Audio/Video Overviews, Mind Maps, Reports incl. Briefing Doc/Study Guide/FAQ, Flashcards, Quiz, slide decks, infographics — discover the exact set from the live Studio panel; the UI evolves fast), and creating new notebooks. Requires browser automation environment — fails gracefully when unavailable.

★ 0
vikasvardhanv
vikasvardhanv/jerry_skills

litreview

Academic literature orientation skill that searches papers via free keyless APIs (PubMed E-utilities + OpenAlex) by default — with the Consensus MCP as an optional enhancement lane when connected — builds a strategic search plan using PICO (default) or SPIDER / Decomposition / hybrid as fallbacks, and synthesizes findings into a formatted Word (.docx) research guide. Grill-me intake (research question specificity + framework hint + tentative depth) before the recon search; a second forcing checkpoint after Phase 2 confirms framework + sub-areas + depth before searches consume budget. Configurable depth (5/10/20 queries) controls coverage vs. speed. Output is a 'launching pad' — an orientation guide that lets a researcher dive in confidently, not a finished review. Use when the user starts literature-oriented research (e.g., 'litreview on [topic]', 'literature review on [topic]', 'I'm starting a literature review on X', 'I'm writing a paper on X', 'help me research X', 'I'm doing research on X', 'can you help me research X'). Do NOT use for single one-off paper searches wanting a quick list — that's a plain PubMed/OpenAlex (or Consensus) query.

★ 0
vikasvardhanv
vikasvardhanv/jerry_skills

grants

NIH grant research skill for clinical researchers. Grill-me intake (research idea + career stage + preliminary data + environment + submission posture + known institute targets) locks down the funding strategy before any search runs. Runs a 5-facet Consensus positioning analysis (with draft Significance/Innovation language), maps the research to the right NIH institutes and study sections via RePORTER, finds NOSIs and funded overlap, and produces an editable Word document (.docx) with budget/scope-aware mechanism recommendations, submission timelines, and a mandatory program officer recommendation. Use when the user asks about research funding or makes any grant-related request (e.g., 'grants for [topic]', 'find grants for my research idea', 'what grants match my research', 'help me find NIH funding', 'grant opportunities for my research'). NIH-only scope — non-NIH funders (PCORI, DOD CDMRP, VA, foundations) are out of scope and flagged at intake.

★ 0