Merge pull request #2 from codeforanchorage/incorporate-upstream-145-144

Adapt upstream enhancements: ACS 1-year hint (uscensusbureau#145) + token flattening (uscensusbureau#144)

Author: brendanbabb

mcp-server/src/tools/fetch-aggregate-data.tool.ts

@@ -20,7 +20,7 @@ import {
   validateGeographyArgs,
 } from '../schema/validators.js'
 
-export const toolDescription = `Fetches Census Bureau statistics for a dataset, vintage, and geography; never guess cell codes -- use search-data-tables first or this tool will reject unknown codes with a "did you mean" hint. Every ACS estimate is auto-paired with its margin of error and flagged LOW RELIABILITY when CV exceeds 30%; suppression sentinels are decoded to text rather than returned as numbers. Required: dataset, year, get (variables or group), and one of for/ucgid. Returns numbered Record blocks with caveats led at the top.`
+export const toolDescription = `Fetches Census Bureau statistics for a dataset, vintage, and geography; never guess cell codes -- use search-data-tables first or this tool will reject unknown codes with a "did you mean" hint. Every ACS estimate is auto-paired with its margin of error and flagged LOW RELIABILITY when CV exceeds 30%; suppression sentinels are decoded to text rather than returned as numbers. ACS 1-year (acs/acs1) only covers areas with populations of 65,000+; use acs/acs5 for smaller geographies or the API returns a bare 400. Required: dataset, year, get (variables or group), and one of for/ucgid. Returns numbered Record blocks with caveats led at the top.`
 
 export class FetchAggregateDataTool extends BaseTool<TableArgs> {
   name = 'fetch-aggregate-data'
@@ -218,6 +218,14 @@ function buildApiErrorMessage(
       `Recover by calling list-datasets to confirm the dataset exists and which vintages are published.`,
     )
   } else if (status === 400) {
+    if (isAcs1YearDataset(ctx.dataset)) {
+      lines.push(
+        `ACS 1-year estimates (acs/acs1) are only published for geographic areas with ` +
+          `populations of 65,000 or more, so a 400 here often means the requested area is too small. ` +
+          `Switch the dataset to acs/acs5 (5-year estimates) to cover smaller geographies. ` +
+          `See https://www.census.gov/programs-surveys/acs/guidance/estimates.html.`,
+      )
+    }
     lines.push(
       `The Census API rejected the query as malformed.`,
       `Re-verify the geography (for/in/ucgid) via fetch-dataset-geography and ` +
@@ -231,6 +239,13 @@ function buildApiErrorMessage(
   return lines.join(' ')
 }
 
+// ACS 1-year estimates are only published for areas of 65,000+ people, a
+// frequent cause of opaque 400s. Detect acs1 regardless of spacing/case so the
+// error path can hand back the population-threshold hint above.
+function isAcs1YearDataset(dataset: string): boolean {
+  return dataset.toLowerCase().replace(/\s+/g, '').includes('acs/acs1')
+}
+
 function buildQueryEcho(opts: {
   dataset: string
   year: number | string

mcp-server/src/tools/fetch-dataset-geography.tool.ts

@@ -218,7 +218,7 @@ export class FetchDatasetGeographyTool extends BaseTool<FetchDatasetGeographyArg
             content: [
               {
                 type: 'text',
-                text: `Available geographies for ${args.dataset}${args.year ? ` (${args.year})` : ''}:\n\n${JSON.stringify(parsedGeographyData, null, 2)}`,
+                text: `Available geographies for ${args.dataset}${args.year ? ` (${args.year})` : ''}:\n\n${JSON.stringify(parsedGeographyData)}`,
               },
             ],
           }

mcp-server/tests/tools/fetch-aggregate-data/fetch-aggregate-data.tool.test.ts

@@ -384,6 +384,32 @@ describe('FetchAggregateDataTool', () => {
       expect(text).toContain('list-datasets')
     })
 
+    it('explains the ACS 1-year 65,000 population threshold on a 400 and suggests acs/acs5', async () => {
+      mockFetch.mockImplementation((url: string) => {
+        if (url.includes('/variables.json')) {
+          return Promise.resolve(new Response('nope', { status: 404 }))
+        }
+        return Promise.resolve(
+          new Response('bad request', {
+            status: 400,
+            statusText: 'Bad Request',
+          }),
+        )
+      })
+
+      const args = {
+        dataset: 'acs/acs1',
+        year: 2022,
+        get: { group: 'B01001' },
+        for: 'place:00100',
+      }
+
+      const response = await tool.toolHandler(args, process.env.CENSUS_API_KEY)
+      const text = response.content[0].text as string
+      expect(text).toContain('65,000')
+      expect(text).toContain('acs/acs5')
+    })
+
     it('should handle network errors', async () => {
       // Variables fetch fails silently; data fetch rejects.
       mockFetch.mockImplementation((url: string) => {