fix(docs): turn off table padding in the markdown serializer and assert exact catalog URLs

Author: waleedlatif1

apps/docs/lib/llms.ts

@@ -4,36 +4,6 @@ import { getIntegrationLandingPage, INTEGRATION_LANDING_LABEL } from '@/lib/inte
 import { getApiSpecContent } from '@/lib/openapi'
 import type { PageData, source } from '@/lib/source'
 
-/** A table row, captured as its indentation (tables nest under list items) and its cells. */
-const TABLE_ROW = /^(\s*)\|(.*)\|$/
-const TABLE_DELIMITER_CELL = /^\s*(:?)-+(:?)\s*$/
-const UNESCAPED_PIPE = /(?<!\\)\|/
-
-/**
- * The processed-markdown serializer pads every table cell to its column's widest cell, so one long
- * enum (a parameter listing every timezone) pads every row of its table and roughly doubles the
- * integration reference. Drop that trailing padding; a cell's own leading indentation (the `↳`
- * nesting in output tables) and inline code are kept, and the tables stay valid GFM. Tables and
- * fences nested under a list item are indented, so both are matched after their indentation.
- */
-function compactTables(markdown: string): string {
-  let inFence = false
-  return markdown
-    .split('\n')
-    .map((line) => {
-      if (line.trimStart().startsWith('```')) inFence = !inFence
-      const row = inFence ? null : TABLE_ROW.exec(line)
-      if (!row) return line
-      const [, indent, inner] = row
-      const cells = inner.split(UNESCAPED_PIPE)
-      if (cells.every((cell) => TABLE_DELIMITER_CELL.test(cell))) {
-        return `${indent}|${cells.map((cell) => cell.replace(TABLE_DELIMITER_CELL, ' $1---$2 ')).join('|')}|`
-      }
-      return `${indent}|${cells.map((cell) => ` ${cell.replace(/^ /, '').trimEnd()} `).join('|')}|`
-    })
-    .join('\n')
-}
-
 type LLMPageData = PageData & { getAPIPageProps?: () => ApiPageProps }
 
 /**
@@ -66,5 +36,5 @@ export async function getLLMText(page: InferPageType<typeof source>) {
     : ''
   return `${heading}
 
-${landingLink}${compactTables(processed)}`
+${landingLink}${processed}`
 }

apps/docs/source.config.ts

@@ -1,5 +1,6 @@
 import { execFileSync } from 'node:child_process'
 import path from 'node:path'
+import { remarkGfm } from 'fumadocs-core/mdx-plugins'
 import { defineConfig, defineDocs, frontmatterSchema } from 'fumadocs-mdx/config'
 import lastModified from 'fumadocs-mdx/plugins/last-modified'
 import { curlJsonBodyGrammar } from './lib/shiki-curl-json'
@@ -66,6 +67,22 @@ export default defineConfig({
     lastModified({ versionControl: async (file) => gitLastModified?.get(path.resolve(file)) }),
   ],
   mdxOptions: {
+    /**
+     * The processed markdown behind `.md` pages, `llms-full.txt`, and Copy page is serialized by
+     * remark-gfm, which by default pads every table cell to its column's widest cell — one long
+     * enum pads a whole table. `tablePipeAlign: false` keeps the table and drops the padding; it
+     * only affects serializing to markdown, never the rendered page.
+     */
+    remarkPlugins: (plugins) => {
+      if (!plugins.includes(remarkGfm)) {
+        throw new Error(
+          'The fumadocs-mdx preset no longer registers remarkGfm; reapply tablePipeAlign'
+        )
+      }
+      return plugins.map((plugin) =>
+        plugin === remarkGfm ? [remarkGfm, { tablePipeAlign: false }] : plugin
+      )
+    },
     /**
      * Shiki defaults to `github-light` / `github-dark`, whose blues and purples appear nowhere
      * in the product. These themes carry the platform's own token colors instead — see

apps/sim/app/.well-known/api-catalog/route.ts

@@ -11,7 +11,8 @@ const API_CATALOG_CONTENT_TYPE =
  * RFC 9727 API catalog: one machine-readable entry point to the v2 REST API, linking its OpenAPI
  * document (`service-desc`), reference docs (`service-doc`), and health endpoint (`status`) with
  * the RFC 8631 relations. The anchor is this deployment's own origin, so a self-hosted instance
- * describes itself; the OpenAPI document is the published contract every deployment serves.
+ * describes itself. The OpenAPI document is the one published contract; its `servers` names Sim
+ * Cloud, so a client of another deployment uses the anchor as its base URL, as the API docs say.
  * `Link: rel="api-catalog"` makes a HEAD request discoverable too (RFC 9727 §2).
  */
 export const GET = withRouteHandler(() => {

apps/sim/scripts/test-mcp-host-e2e.ts

@@ -3,6 +3,7 @@ import { mkdir, writeFile } from 'node:fs/promises'
 import { dirname } from 'node:path'
 import { createLogger } from '@sim/logger'
 import { getErrorMessage } from '@sim/utils/errors'
+import { SIM_DOCS_URL } from '@sim/utils/site'
 
 /**
  * Exercises the public ownership challenge, host isolation, the RFC 9727 API catalog, and the
@@ -117,15 +118,15 @@ try {
       const { response, body } = await request(API_CATALOG_PATH, userAgent, baseUrl.host)
       assert.equal(response.status, 200)
       assert.equal(response.headers.get('content-type'), API_CATALOG_CONTENT_TYPE)
-      assert.match(
-        response.headers.get('link') ?? '',
-        /\/\.well-known\/api-catalog>; rel="api-catalog"$/
+      assert.equal(
+        response.headers.get('link'),
+        `<${baseUrl.origin}${API_CATALOG_PATH}>; rel="api-catalog"`
       )
       const [api] = JSON.parse(body).linkset
-      assert.equal(new URL(api.anchor).pathname, '/api/v2')
-      assert.equal(new URL(api['service-desc'][0].href).pathname, '/openapi.json')
-      const status = new URL(api.status[0].href)
-      const health = await request(status.pathname, 'Mozilla/5.0', baseUrl.host)
+      assert.equal(api.anchor, `${baseUrl.origin}/api/v2`)
+      assert.equal(api['service-desc'][0].href, `${SIM_DOCS_URL}/openapi.json`)
+      assert.equal(api.status[0].href, `${baseUrl.origin}/api/health`)
+      const health = await request(new URL(api.status[0].href).pathname, userAgent, baseUrl.host)
       assert.equal(health.response.status, 200, 'The catalog status link must resolve')
     })
   }