Atlas Sentinel

v1.0.0

A powerful MCP server built with NitroStack

Connection Setup

Add via Cursor Settings UI (Settings > Features > MCP > Add New MCP Server):

{
  "mcpServers": {
    // your other mcp servers
    "atlas-sentinel": {
      "url": "https://atlas-sentine-zero-friction-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"
    }
  }
}

Connect remote tools directly via Claude's Web UI:

Add custom connector BETA
Connect Claude to your data and tools. Learn more about connectors or get started with pre-built ones.
Advanced settings
Only use connectors from developers you trust. Anthropic does not control which tools developers make available and cannot verify that they will work as intended or that they won't change.

Configure custom tools directly via ChatGPT's Web UI:

New App
PNG only. Best results at 256 x 256 px or larger. Max file size: 10 KB
Custom MCP servers introduce risk. Learn more
OpenAI hasn't reviewed this MCP server. Attackers may attempt to steal your data or trick the model into taking unintended actions, including destroying data.

Add the following configuration block under mcpServers in your Antigravity configuration file (~/.gemini/config/mcp_config.json):

{
  "mcpServers": {
    // your other mcp servers
    "atlas-sentinel": {
      "serverUrl": "https://atlas-sentine-zero-friction-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"
    }
  }
}

Add the following configuration block to your Codex configuration file (~/.codex/config.toml):

[mcp_servers.atlas-sentinel]
url = "https://atlas-sentine-zero-friction-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp"

Connect directly using the Server-Sent Events endpoint:

https://atlas-sentine-zero-friction-amrita-university-amritapuri-campus.app.nitrocloud.ai/mcp
Available Tools
register_asset

Register a physical asset (factory, warehouse, supplier, office, datacenter, port) that Atlas should monitor for threats. Use whenever the user mentions a facility, plant, supplier or site they care about. Requires a name and coordinates; if the user gives only a city, estimate its lat/lon yourself and pass them in.

list_assets

List every asset currently monitored by Atlas, with coordinates and type. Use to answer "what am I monitoring?" or before threat assessments when the user refers to "my assets/factories/suppliers".

remove_item

Stop monitoring an asset (kind='asset') or delete an alert contact (kind='contact'), by exact name. Use when the user asks to remove/deregister a site or a contact.

latest_events

Fetch real recent earthquakes from the USGS global feed, with magnitude, location, coordinates, hours ago, and a unique event id per quake. WHEN TO USE: any question like "Any significant earthquakes in the last day?", "any earthquakes?", "recent seismic activity?" — this is always the FIRST tool of the earthquake workflow. If the user also asks "are any of my monitored assets at risk?" or "are my factories affected?", follow up with check_asset_exposure using an id from events[].id. WHEN NOT TO USE: historical/famous earthquakes (replay_event); a full multi-channel risk overview (threat_sweep).

check_asset_exposure

THE core risk tool: given an earthquake event id (from latest_events or replay_event), compute how exposed each monitored asset is, using distance from the epicenter and magnitude. Severity levels: severe / high / moderate / low / none. Use whenever the user asks "are we affected", "which sites are at risk", or after any significant event is found.

find_critical_infra

Find critical infrastructure (hospitals, fire stations) near a coordinate via OpenStreetMap. Use when assessing an earthquake epicenter or asset location — e.g. for a situation report listing nearby emergency facilities.

replay_event

SIMULATION MODE: run the full Atlas exposure pipeline against a famous historical earthquake, clearly labeled as a replay. Use for capability demonstrations ("show me how Atlas would respond to the 2023 Türkiye earthquake"). Known ids: "us6000jllz" = 2023 Türkiye M7.8. Any USGS event id works.

forecast_at

Get the weather over the next few days (3-day severe-weather outlook) for ONE monitored asset: storm risk, flood risk, typhoon or damaging wind gusts, extreme rainfall, and heatwaves. WHEN TO USE: any question about weather risk, storm risk, flood risk, typhoons, wind, rain or heat at a specific site — e.g. "What is the storm and flood risk for my Osaka Distribution Warehouse over the next few days?". Call this tool DIRECTLY with the asset name the user gave — do NOT call list_assets first. Matching is case-insensitive and a unique partial name (e.g. "Osaka") also works; an unknown name returns known_assets so you can recover in one step. WHEN NOT TO USE: weather across all assets at once (threat_sweep); earthquake exposure (check_asset_exposure).

space_weather

Current geomagnetic storm conditions from NOAA (planetary K-index). Solar storms disrupt GPS, satellite links and power grids — relevant to logistics, datacenters and grid-dependent factories. Use when the user asks about solar activity, GPS reliability, or as part of a full threat sweep.

news_shocks

Scan global news (GDELT, updated every 15 min) for supply-chain disruption signals: strikes, port closures, accidents, unrest. Use with a focused query like "port strike Taiwan" or "factory fire semiconductor". Treat results as SIGNALS to investigate, not confirmed facts, and say so.

threat_sweep

THE flagship tool: sweep all four hazard domains in parallel — real USGS earthquakes, 3-day severe weather per asset, NOAA geomagnetic storms, and GDELT news signals per asset region — and return scored raw SIGNALS. WHEN TO USE: "give me a morning brief", "what threatens my operations right now", "risk overview across all hazard channels", "current threats", operations or supply-chain risk questions — any request for an overall picture. It deliberately does NOT return a final verdict: YOU must correlate signals across channels (e.g. a quake near an asset PLUS a disruption headline from the same region is worse than either alone) and produce your own ranked judgment. The response is a compact summary; read resource atlas://threats/live for the complete signal set before deep analysis. WHEN NOT TO USE: one specific earthquake (check_asset_exposure); weather at one named site (forecast_at).

generate_sitrep

Generate a formal situation report for one earthquake event id (from latest_events or replay_event): event facts, per-asset exposure, severity-matched recommended actions. WHEN TO USE: "build me a formal situation report", "a sitrep I can send to management", "a report to circulate" — any request for a formal, shareable earthquake report. Pass the event id the user gave or one returned by a prior tool. Returns a summary; the full report is stored at resource atlas://sitrep/latest. WHEN NOT TO USE: multi-channel overviews (threat_sweep); reports without an earthquake event id.

register_contact

Register a stakeholder to receive Telegram alerts for a specific asset. IMPORTANT: the contact must first open Telegram, find the bot, and press Start — Telegram does not allow bots to message users who have not initiated contact.

list_contacts

List all registered alert contacts with their assets and roles.

send_alert

Send a Telegram alert to registered contacts (optionally only those of one asset), OR to one specific telegram_chat_id. GUARDRAIL: direct chat ids must already be in the contacts registry or equal the server owner id (prevents spam relay). GUARDRAIL: first call WITHOUT confirm returns a preview — show it to the user and only re-call with confirm=true after explicit approval. Use when the user says 'notify the team' or 'alert the factory owner'.

auto_alert_sweep

Run a fresh seismic exposure check and AUTOMATICALLY notify registered contacts via Telegram for any threat at or above the configured threshold (no confirmation needed — this is the autonomous path, governed by budget/rate/kill-switch guardrails). Use when the user says 'check and alert', 'run alerts', or 'notify if anything is wrong'.

monitoring

Control the autonomous background watchman with one tool. action='start' (with optional interval_minutes, default 5): Atlas checks USGS every N minutes and auto-alerts contacts above the threshold even when nobody is chatting — use when the user says 'watch my assets' or 'alert me if an earthquake happens'. action='stop' halts it. action='status' reports active state and statistics. Note: the timer runs while the server process is alive; serverless scale-to-zero pauses it when the server sleeps.

alert_controls

Configure the alerting guardrails with one tool. action='set_threshold' + level (severe/high/moderate/low): minimum severity that triggers alerts — use for 'alert me only for severe' or 'notify on moderate or above'. action='pause': KILL SWITCH, instantly stop ALL sends (use immediately if the user reports spam or wants silence). action='resume': re-enable. action='set_budget' + per_day: max Telegram alerts per day (default 20). action='status': current threshold, pause state, budget remaining.