@cyanheads/fema-mcp-server
Query FEMA disaster declarations, public assistance grants, housing aid, and NFIP flood insurance claims via MCP. STDIO or Streamable HTTP.
8 Tools • 1 Opt-in Tool • 1 Resource • 1 Skill
[](./CHANGELOG.md) [](./LICENSE) [](https://github.com/users/cyanheads/packages/container/package/fema-mcp-server) [](https://modelcontextprotocol.io/) [](https://www.npmjs.com/package/@cyanheads/fema-mcp-server) [](https://www.typescriptlang.org/) [](https://bun.sh/)
[](https://github.com/cyanheads/fema-mcp-server/releases/latest/download/fema-mcp-server.mcpb) [](https://cursor.com/en/install-mcp?name=fema-mcp-server&config=eyJjb21tYW5kIjoibnB4IiwiYXJncyI6WyIteSIsIkBjeWFuaGVhZHMvZmVtYS1tY3Atc2VydmVyIl19) [](https://vscode.dev/redirect?url=vscode:mcp/install?%7B%22name%22%3A%22fema-mcp-server%22%2C%22command%22%3A%22npx%22%2C%22args%22%3A%5B%22-y%22%2C%22%40cyanheads%2Ffema-mcp-server%22%5D%7D)
[](https://www.npmjs.com/package/@cyanheads/mcp-ts-core)
**Public Hosted Server:** [https://fema.caseyjhand.com/mcp](https://fema.caseyjhand.com/mcp)
---
## Overview
Federal disaster recovery data from FEMA: disaster declarations, public assistance grants, individual housing assistance, and NFIP flood insurance claims. Search declarations by state or incident, see the public and housing assistance behind any disaster, and run SQL over large flood-claim result sets.
### Tools
| Tool | Description |
|:---|:---|
| `fema_search_disasters` | Search federal disaster declarations by title, state, incident type, declaration type, date range, and county |
| `fema_get_disaster` | Fetch up to 1,000 designated-area records with per-area program flags and partial-coverage disclosure |
| `fema_get_public_assistance` | Public assistance funded projects for a disaster or state — where federal recovery money went |
| `fema_get_housing_assistance` | Individual assistance housing data for a disaster — owner and renter breakdowns and inspection counts by county/ZIP, filterable by county or ZIP |
| `fema_search_nfip` | NFIP flood insurance claims for a state, county, or ZIP, with DataCanvas spillover for SQL analytics |
| `fema_dataframe_describe` | List columns and row counts for DataCanvas tables staged by `fema_search_nfip` |
| `fema_dataframe_query` | Run a SELECT query against a DataCanvas table staged by `fema_search_nfip` |
| `fema_dataframe_drop` | Remove a staged DataCanvas table or view — opt-in, absent from `tools/list` by default |
| `fema_query_dataset` | Generic OData query against any dataset in the OpenFEMA catalog — escape hatch for datasets the convenience tools don't cover |
`fema_dataframe_drop` is opt-in: it appears in `tools/list` only when `FEMA_ENABLE_CANVAS_DROP=true`. The other eight tools are always listed.
### Resources
| Resource | Description |
|:---|:---|
| `fema://disaster/{disasterNumber}` | Summary for a specific FEMA disaster declaration — title, state, incident type, programs, incident period |
All resource data is also reachable via tools. Use `fema_get_disaster` for designated-area detail (up to 1,000 rows).
### Skills
| Skill | Description |
|:---|:---|
| `fema-guide` | Gates, identifier sources, and OData syntax for the `fema_*` tools: county naming across datasets, dollar totals, NFIP scoping, and `fema_query_dataset` filters |
Plugin installs pick up the root `skills/` tree; MCP clients that support the Skills extension read it as `skill://fema-guide/SKILL.md`.
## Capability reference
### `fema_search_disasters` tool
- Filters: `title`, `state`, `incident_type`, `declaration_type` (`DR`/`EM`/`FM`), inclusive `date_from`/`date_to` (`YYYY-MM-DD`), and `county`. `title`, `incident_type`, and `county` are case-insensitive substring matches. A title such as `harvey` also matches unrelated names such as HARVEYTOWN FIRE. Page with `limit` (1–1000, default 50) and `offset`; a page stops at 100,000 characters and `next_offset` says where to resume.
- Returns one summary per disaster number with `designated_area_count`, program flags, and declaration totals. With `county`, `total_area_rows` and `matched_areas` (FIPS codes and per-area flags) show the matching areas. Errors: `invalid_state`, `invalid_range` (`date_from` after `date_to`), `request_too_long`.
- Searches the most recent 10,000 designated-area rows. When capped, results are marked `truncated`, totals are lower bounds, and `notice` names the next `date_to`.
County filters across declarations, housing, and PA recognize terminal County, Parish, Borough, and Census Area forms with or without parentheses (`Harris County` and `Harris (County)`). Recognized names use the stored counterpart spelling; unknown names retain exact/case fallbacks. Bare `harris` still matches Harrison; a suffixed name retains its type. IA flags combine legacy IA and IHP declarations; they do not guarantee housing records.
---
### `fema_get_disaster` tool
- Requires `disaster_number` from `fema_search_disasters`.
- Returns up to 1,000 designated-area records with FIPS codes and each area's `ia_declared`, `pa_declared`, and `hm_declared` flags, incident period, and program flags combined across fetched areas; a missing disaster returns `not_found`. When more rows exist, `areas_truncated`, `total_designated_area_count`, and `notice` disclose partial coverage. False flags do not exclude programs in unfetched rows; use `fema_query_dataset` for the full declaration.
---
### `fema_get_public_assistance` tool
- Requires `disaster_number` or `state`; optional `county` (case-insensitive substring), `limit` (1–10000, default 100), and `offset`. A page stops at 100,000 characters; `next_offset` says where to resume.
- Returns applicants, damage categories, project size/status, and federal and total obligations. Errors: `invalid_state`, `missing_filter`, `not_found`, `request_too_long`.
---
### `fema_get_housing_assistance` tool
- Requires `disaster_number`; optional `state`, `county` (case-insensitive substring), `zip_code` (5 digits), `type` (`owners`/`renters`/`both`, default `both`), `limit` (1–10000, default 100), and `offset`. A page stops at 100,000 characters; one `next_offset` continues both owners and renters.
- Returns separate `owners` and `renters` arrays with county/ZIP registrations, approved amounts, and inspection counts, plus `owners_count`/`renters_count`. Owners also carry inspected damage in USD; renters carry damage-degree counts only. Errors: `invalid_state`, `not_found`, `request_too_long`.
---
### `fema_search_nfip` tool
- Requires `state`; optional `county_code` (5-digit FIPS like `48201`, or a 3-digit code like `201` that gets the state prefix), `zip_code`, `year_from`/`year_to`, `limit` (1–10000, default 1000), and `offset`. Results come newest loss first.
- Inline claims stop at 100,000 characters; `next_offset` and `notice` say how to continue. Errors: `invalid_state`, `invalid_county_code` (anything but 1–5 digits, or a short code for FM, MH, or PW), `invalid_range` (`year_from` after `year_to`).
- County searches check for claims without a county code across the same state/year range, using one count-only exchange with no retries and a two-second deadline. The response uses the count only if ready with the primary page; otherwise it says coverage was not checked. These claims cannot be allocated to the selected county or ZIP. Removing `county_code` and optionally searching known ZIPs may help; ZIP results are not guaranteed complete or equivalent to a county.
- With DataCanvas on (the default), a first-page match over the inline limit stages up to 50,000 rows. `canvas_id`, `canvas_table`, and `staged_count` identify the stage, `truncated: true` marks its cap, and `total_count` is the full match. Inspect it with `fema_dataframe_describe`, then query it with `fema_dataframe_query`.
---
### `fema_dataframe_describe` tool
- Requires `canvas_id` from `fema_search_nfip`.
- Lists table/view names, column types, nullability, and staged row counts. Typed errors: `canvas_not_found`, `canvas_unavailable`.
---
### `fema_dataframe_query` tool
- Requires `canvas_id` and a single SQL `query`; only SELECT is allowed.
- Rows are capped at the canvas row limit and 100,000 characters; `truncated`/`shown` and OFFSET guidance appear when the result is cut. Errors: `canvas_unavailable`, `canvas_not_found`, `missing_table`, `invalid_sql`, `non_select_statement`, `multi_statement`, `denied_function`, `plan_operator_not_allowed`, `sql_read_only`, `sql_execution_error` (use `TRY_CAST` or filter incompatible values).
---
### `fema_dataframe_drop` tool
- Requires `canvas_id` and the exact `table_name` from `fema_dataframe_describe`.
- Removes one staged table or view; `dropped: false` means the name did not exist. Errors: `canvas_not_found`, `canvas_unavailable`.
- Opt-in through `FEMA_ENABLE_CANVAS_DROP=true`.
---
### `fema_query_dataset` tool
- Requires a case-sensitive OpenFEMA `dataset`; optional OData `filter`/`select`/`orderby`, `limit` (1–10000, default 100), and `offset`. A page stops at 100,000 characters; `next_offset` says where to resume.
- Returns records and `total_count`. Errors: `unknown_dataset`, `dataset_not_served`, `catalog_unavailable` (retryable), `invalid_filter`, `invalid_select`, `invalid_orderby`, `invalid_odata_syntax`, `request_too_long` (URL past about 8,900 characters), `filter_too_complex` (nested more than 100 levels).
- For a list of values, use `field in ('a','b')` instead of chained `or` comparisons (about 90 allowed). For `NfipPolicies`, use `propertyState` and `reportedZipCode`, and always include a ZIP filter to avoid timeout.
---
### `fema://disaster/{disasterNumber}` resource
- Requires `disasterNumber` as a string of digits.
- Returns title, state, incident type, declaration type/date, incident period, `programs_declared`, and fetched `designated_area_count` (up to 500). When more rows exist, the summary includes `areas_truncated`, `total_designated_area_count`, and `notice`: programs describe only fetched rows, and an empty list does not establish absence. Use `fema_query_dataset` for full rows. `not_found` points to `fema_search_disasters`.
## Features
Built on [`@cyanheads/mcp-ts-core`](https://github.com/cyanheads/mcp-ts-core): stdio and Streamable HTTP transports, pluggable auth (`none` / `jwt` / `oauth`), swappable storage (`in-memory`, `filesystem`, `Supabase`, `Cloudflare KV/R2/D1`), structured logging with optional OpenTelemetry tracing.
OpenFEMA-specific:
- Case-insensitive county and incident type filters, using bundled declaration names plus exact PA county spellings. Refresh with `bun run vocabulary:build`: sequential full exports read declarations in ID order and nonnull PA counties without a sort, with bounded transient retries. Each export must match its reported count; a failed or incomplete refresh leaves the snapshot unchanged.
- `DisasterDeclarationsSummaries` rows collapse to one summary per declaration, with `designated_area_count`
- `fema_search_nfip` requires `state`, which prevents unbounded 2.7M-row fetches
- Large NFIP result sets spill into DuckDB tables you can query with SQL via DataCanvas
- No API keys required — OpenFEMA is a free, public API
Agent-friendly output:
- Disaster number is the join key across tools, so calls chain without re-searching
- Empty searches return empty arrays, zero totals, and a `notice` with guidance for supplied filters on both response surfaces; assistance calls can check declared programs with `fema_get_disaster`. Paging past a nonzero total has separate offset guidance.
- Typed errors on every tool (`invalid_state`, `invalid_county_code`, `invalid_range`, `not_found`, `missing_filter`, `unknown_dataset`, `dataset_not_served`, `invalid_filter`, `invalid_select`, `invalid_orderby`, `invalid_odata_syntax`, `request_too_long`, `filter_too_complex`, `canvas_unavailable`), each with a recovery hint
- `designated_area_count` on search results shows whether `fema_get_disaster` is worth a follow-up
## Getting started
Connect to the hosted instance, no install:
```sh
claude mcp add --transport http fema-mcp-server https://fema.caseyjhand.com/mcp
```
Or run it locally:
```sh
claude mcp add fema-mcp-server -- npx -y @cyanheads/fema-mcp-server
```
For other clients, add this to the MCP config file:
```json
{
"mcpServers": {
"fema-mcp-server": {
"command": "npx",
"args": ["-y", "@cyanheads/fema-mcp-server"]
}
}
}
```
No API key is required — OpenFEMA is a free, public API. For Docker or a source build, see [Running the server](#running-the-server).
## Configuration
All configuration is validated at startup via Zod schemas in `src/config/server-config.ts`. Key environment variables:
| Variable | Description | Default |
|:---------|:------------|:--------|
| `FEMA_BASE_URL` | OpenFEMA API root. | `https://www.fema.gov/api/open` |
| `FEMA_REQUEST_TIMEOUT_MS` | Total time budget in milliseconds for one upstream request, including retries. | `30000` |
| `CANVAS_PROVIDER_TYPE` | DataCanvas SQL analytics over large NFIP Claims result sets: `duckdb`, or `none` to disable it. The Claude Desktop bundle runs with `none` and returns NFIP claims inline. | `duckdb` |
| `FEMA_ENABLE_CANVAS_DROP` | Enable the `fema_dataframe_drop` tool. | `false` |
| `MCP_TRANSPORT_TYPE` | Transport: `stdio` or `http`. | `stdio` |
| `MCP_SESSION_MODE` | Session mode: `auto`, `stateful`, or `stateless`. | `stateless` |
| `MCP_HTTP_PORT` | HTTP server port. | `3010` |
| `MCP_AUTH_MODE` | Auth mode: `none`, `jwt`, or `oauth`. | `none` |
| `MCP_LOG_LEVEL` | Log level (`debug`, `info`, `warning`, `error`, etc.). | `info` |
| `LOG_TOOL_FAILURE_PAYLOADS` | Log failed-call arguments and results, redacted by key name. Secrets in free-form values are retained. | `false` |
| `LOG_TOOL_FAILURE_PAYLOAD_MAX_BYTES` | Byte cap per logged payload. | `16384` |
| `OTEL_EXPORTER_OTLP_ENDPOINT` | OTLP base URL for traces and metrics. | — |
| `OTEL_EXPORTER_OTLP_TRACES_ENDPOINT` | Traces endpoint, used as-is. | — |
| `OTEL_EXPORTER_OTLP_METRICS_ENDPOINT` | Metrics endpoint, used as-is. | — |
| `OTEL_EXPORTER_OTLP_LOGS_ENDPOINT` | Log export endpoint; logs export only when this is set. | — |
| `LOGS_DIR` | Directory for log files (Node.js only). | `/logs` |
| `OTEL_ENABLED` | Enable [OpenTelemetry instrumentation](https://github.com/cyanheads/mcp-ts-core/tree/main/docs/telemetry). | `false` |
See [`.env.example`](./.env.example) for the full list of optional overrides.
## Running the server
### Local development
Requires [Bun v1.4.0](https://bun.sh/) or higher (or Node.js v24+).
- **Clone and install:**
```sh
git clone https://github.com/cyanheads/fema-mcp-server.git
cd fema-mcp-server && bun install
cp .env.example .env # no required variables
```
- **Build and run:**
```sh
# One-time build
bun run rebuild
# Run the built server
bun run start:stdio
# or
bun run start:http
```
- **Run checks and tests:**
```sh
bun run devcheck # Lint, format, typecheck, security
bun run test # Vitest test suite
bun run lint:mcp # Validate MCP definitions against spec
```
### Docker
```sh
docker run --rm -p 3010:3010 ghcr.io/cyanheads/fema-mcp-server:latest
```
The image defaults to HTTP transport, stateless session mode, and logs to `/var/log/fema-mcp-server`. To run it over stdio from an MCP client, use `"command": "docker"` with `"args": ["run", "-i", "--rm", "-e", "MCP_TRANSPORT_TYPE=stdio", "ghcr.io/cyanheads/fema-mcp-server:latest"]`. To build locally, `docker build -t fema-mcp-server .`; OpenTelemetry peer dependencies are installed by default — build with `--build-arg OTEL_ENABLED=false` to omit them.
## Project structure
| Directory | Purpose |
|:----------|:--------|
| `src/index.ts` | `createApp()` entry point — registers tools/resources and inits services. |
| `src/config` | Server-specific environment variable parsing and validation with Zod. |
| `src/mcp-server/tools` | Tool definitions (`*.tool.ts`) — one file per tool. |
| `src/mcp-server/resources` | Resource definitions (`*.resource.ts`). |
| `src/services/openfema` | OpenFEMA REST API client, plus the case-insensitive name matcher and its bundled name vocabulary. |
| `src/services/canvas` | DataCanvas integration — DuckDB spillover for large NFIP result sets. |
| `skills/` | Published agent skill (`fema-guide`), served over MCP via `createApp({ skills })`. |
| `tests/` | Unit and integration tests mirroring `src/`. |
## Development guide
See [`CLAUDE.md`](./CLAUDE.md) for development guidelines and architectural rules. The short version:
- Handlers throw, framework catches — no `try/catch` in tool logic
- Use `ctx.log` for request-scoped logging, `ctx.state` for tenant-scoped storage
- Register new tools and resources in `src/index.ts`
- Wrap external API calls: validate raw → normalize to domain type → return output schema; never fabricate missing fields
## Contributing
Issues are welcome. Run checks and tests before submitting:
```sh
bun run devcheck
bun run test
```
## Data source
Data is provided by the [OpenFEMA API](https://www.fema.gov/about/openfema/api), a free public API maintained by the Federal Emergency Management Agency. Usage is subject to [OpenFEMA Terms and Conditions](https://www.fema.gov/about/openfema/terms-conditions).
> This product uses the Federal Emergency Management Agency's OpenFEMA API, but is not endorsed by FEMA. The Federal Government or FEMA cannot vouch for the data or analyses derived from these data after the data have been retrieved from the Agency's website(s).
When citing OpenFEMA datasets in research or publications, use the format FEMA specifies:
```
Federal Emergency Management Agency (FEMA), OpenFEMA Dataset: . Retrieved from on .
```
## License
Apache-2.0 — see [LICENSE](./LICENSE) for details.