Ph Civic Data MCP

What xmpuspus/ph-civic-data-mcp MCP server does

The xmpuspus/ph-civic-data-mcp MCP server makes Philippine public data available as MCP tools over a local stdio process. Its coverage includes earthquakes, volcano alert levels, weather forecasts, tropical cyclones, PAGASA alerts, procurement notices, infrastructure projects, population, poverty, inflation, labor, health, administrative locations, air quality, climate, vegetation, and macroeconomic indicators.

The server is suited to questions that require combining sources rather than reading one isolated dataset. For example, get_area_profile resolves a place to a PSGC code and combines demographic, economic, procurement, hazard, and weather information. assess_area_risk combines PHIVOLCS and PAGASA results, while flag_infra_anomalies applies explicit heuristics to infrastructure notices and includes disclaimers that the results are review indicators, not accusations.

How it works

The xmpuspus/ph-civic-data-mcp MCP server retrieves data from public portals and upstream feeds with different formats, including HTML pages, PSA OpenSTAT data, and public APIs. Several results are cached. Procurement searches use roughly the latest 100 PhilGEPS bid notices and filter them in memory; the stated cache period is six hours. The freshness tool reports the running version, source catalog, expected freshness, cache TTLs, licenses, and related metadata.

Location tools support fuzzy place-name resolution, browsing administrative children, and traversing a full region-to-barangay hierarchy. PSA catalog browsing helps an agent discover statistical tables instead of relying only on predefined table identifiers. Environmental tools retrieve NASA POWER climate data, NASA MODIS vegetation indices, and air-quality readings from Open-Meteo.

Some upstream behavior is handled explicitly. Weather forecasting uses the PAGASA TenDay API when PAGASA_API_TOKEN is set and otherwise uses Open-Meteo. Active-typhoon results distinguish an empty result from an upstream error. Weather-alert extraction returns a caveat when the PAGASA page is reachable but its warning state cannot be isolated reliably.

Setup and configuration

Install the package with uvx ph-civic-data-mcp and configure the MCP client to run it over stdio. The README provides configurations for Claude Desktop, Cursor, VS Code, Claude Code, Codex, and Docker. No API key is required for the standard installation. Set PAGASA_API_TOKEN if the PAGASA-backed forecast path is needed; without it, the forecast tool uses Open-Meteo.

The package is published for Python 3.11 and later versions listed by the project. A Dockerfile is also provided for users who want to build and run the server in a container, using interactive stdin for the stdio transport.

Tools and capabilities

The xmpuspus/ph-civic-data-mcp MCP server includes capabilities for:

• Reading PHIVOLCS earthquakes, bulletins, and volcano statuses.
• Checking Philippine forecasts, typhoons, and weather-alert states.
• Searching and summarizing PhilGEPS procurement and infrastructure notices.
• Retrieving PSA population, poverty, inflation, labor, and health indicators.
• Resolving PSGC locations and browsing administrative hierarchies.
• Combining local demographics, procurement, hazards, and weather in an area profile.
• Querying solar, climate, air-quality, vegetation, USGS earthquake, historical typhoon, and World Bank data.
• Browsing one level of the PSA OpenSTAT catalog.

Limitations and notes

The xmpuspus/ph-civic-data-mcp MCP server depends on public upstream sources, so availability, publication schedules, and freshness vary. PSA values may refer to the latest published period rather than the current month. The national labor indicator series does not provide regional breakdowns; a supplied region is recorded as a caveat.

PhilGEPS open notices do not generally publish approved budget amounts or project-progress data. Consequently, cost_php is often null, and the high-cost anomaly rule is a threshold-based transparency flag rather than evidence that progress is missing. Hazard-overlap and duplicate-title flags also require independent review. PAGASA weather-alert parsing deliberately avoids fabricating advisories when page content is ambiguous, and the project directs users to the PAGASA site for real-time advisories.