fix(exa): keep legacy research model, add subblock migrations, sharpen outputs

The subblock ID stability check caught the removed subblocks — that gate
exists precisely to stop removals from breaking deployed workflows.

- Restore the research model sub-block, scoped to the legacy exa_research
  operation so it never shows for new workflows but still serializes for
  saved ones, and restore the model to effort mapping. Removing it lost
  the configured research depth, silently falling back to effort auto.
- Register useAutoprompt and livecrawl in SUBBLOCK_ID_MIGRATIONS as
  intentional removals. Neither has a value-compatible replacement:
  livecrawl is a mode string and maxAgeHours a number, so mapping one to
  the other would send NaN.
- Replace vague json output descriptions with their inner field lists.
- Drop the separate compat test file; the coverage that guards real
  regressions now lives in exa.test.ts.

Author: waleedlatif1

apps/sim/blocks/blocks/exa.ts

@@ -25,6 +25,16 @@ const CATEGORY_OPTIONS = [
 const LEGACY_RESEARCH_OPERATION = 'exa_research'
 const AGENT_OPERATIONS = ['exa_agent', LEGACY_RESEARCH_OPERATION]
 
+/**
+ * Maps the retired research models onto the Agent API's effort levels, so a
+ * workflow that asked for a deeper (or cheaper) run still gets one.
+ */
+const RESEARCH_MODEL_TO_EFFORT: Record<string, string> = {
+  'exa-research-fast': 'low',
+  'exa-research': 'medium',
+  'exa-research-pro': 'high',
+}
+
 export const ExaBlock: BlockConfig<ExaResponse> = {
   type: 'exa',
   name: 'Exa',
@@ -411,6 +421,18 @@ export const ExaBlock: BlockConfig<ExaResponse> = {
       condition: { field: 'operation', value: AGENT_OPERATIONS },
       mode: 'advanced',
     },
+    {
+      id: 'model',
+      title: 'Research Model (Legacy)',
+      type: 'dropdown',
+      options: [
+        { label: 'Standard', id: 'exa-research' },
+        { label: 'Fast', id: 'exa-research-fast' },
+        { label: 'Pro', id: 'exa-research-pro' },
+      ],
+      description: 'Retired Exa research model, carried over as an Agent effort level',
+      condition: { field: 'operation', value: LEGACY_RESEARCH_OPERATION },
+    },
     {
       id: 'previousRunId',
       title: 'Previous Run ID',
@@ -556,6 +578,10 @@ export const ExaBlock: BlockConfig<ExaResponse> = {
         if (params.livecrawlTimeout) {
           result.livecrawlTimeout = Number(params.livecrawlTimeout)
         }
+        /** Carry a retired research model over to the Agent API's effort scale. */
+        if (params.operation === LEGACY_RESEARCH_OPERATION && params.model) {
+          result.effort = RESEARCH_MODEL_TO_EFFORT[params.model as string] ?? 'medium'
+        }
         return result
       },
     },
@@ -595,20 +621,43 @@ export const ExaBlock: BlockConfig<ExaResponse> = {
     excludeSourceDomain: { type: 'boolean', description: 'Exclude source domain' },
     // Agent operation
     effort: { type: 'string', description: 'Agent effort level' },
+    model: { type: 'string', description: 'Retired research model, mapped to an effort level' },
     previousRunId: { type: 'string', description: 'Agent run to continue from' },
   },
   outputs: {
     // Search and Get Contents output
-    results: { type: 'json', description: 'Search or content results' },
-    statuses: { type: 'json', description: 'Per-URL crawl outcome for Get Contents' },
-    structuredOutput: { type: 'json', description: 'Synthesized output matching the schema' },
-    grounding: { type: 'json', description: 'Field-level citations for generated output' },
-    requestId: { type: 'string', description: 'Exa request identifier' },
+    results: {
+      type: 'json',
+      description:
+        '[{id, title, url, publishedDate, author, summary, favicon, image, text, highlights, highlightScores, subpages, entities, extras}]',
+    },
+    statuses: {
+      type: 'json',
+      description: 'Get Contents only. [{id, status, source, error}] — per-URL crawl outcome',
+    },
+    structuredOutput: {
+      type: 'json',
+      description: 'Search only. Synthesized output matching the supplied output schema',
+    },
+    grounding: {
+      type: 'json',
+      description: '[{field, citations, confidence}] — field-level citations for generated output',
+    },
+    requestId: { type: 'string', description: 'Exa request identifier, useful for support' },
     // Find Similar Links output
-    similarLinks: { type: 'json', description: 'Similar links found' },
+    similarLinks: {
+      type: 'json',
+      description: '[{id, title, url, text, summary, highlights, score}]',
+    },
     // Answer output
-    answer: { type: 'json', description: 'Generated answer' },
-    citations: { type: 'json', description: 'Answer citations' },
+    answer: {
+      type: 'json',
+      description: 'Generated answer — a string, or an object when an output schema is supplied',
+    },
+    citations: {
+      type: 'json',
+      description: '[{id, title, url, text, author, publishedDate}]',
+    },
     // Agent output
     runId: { type: 'string', description: 'Agent run identifier' },
     status: { type: 'string', description: 'Agent run status' },
@@ -617,7 +666,8 @@ export const ExaBlock: BlockConfig<ExaResponse> = {
     structured: { type: 'json', description: 'Agent structured result' },
     research: {
       type: 'json',
-      description: 'Agent answer in the retired Research operation output shape',
+      description:
+        '[{title, url, summary, text, score}] — the agent answer in the retired Research operation shape, so saved workflows keep resolving',
     },
   },
 }

apps/sim/lib/workflows/migrations/subblock-migrations.ts

@@ -65,6 +65,18 @@ export const SUBBLOCK_ID_MIGRATIONS: Record<string, Record<string, string>> = {
     stage_ids: '_removed_stage_ids',
     owner_ids: '_removed_owner_ids',
   },
+  exa: {
+    /**
+     * Exa deprecated both fields. `useAutoprompt` is gone from the API, and
+     * `livecrawl` is superseded by `maxAgeHours` — but their values are not
+     * interchangeable (`livecrawl` is a mode string, `maxAgeHours` a number),
+     * so mapping one onto the other would send `NaN`. Dropping `livecrawl` is
+     * also the fix for the block having defaulted it to `never`, which pinned
+     * every saved search to cached results.
+     */
+    useAutoprompt: '_removed_useAutoprompt',
+    livecrawl: '_removed_livecrawl',
+  },
   rippling: {
     action: '_removed_action',
     candidateDepartment: '_removed_candidateDepartment',

apps/sim/tools/exa/exa-compat.test.ts

@@ -170,6 +170,27 @@ describe('exa block', () => {
     expect(ExaBlock.tools.config?.tool?.({ operation: 'exa_research' })).toBe('exa_agent')
   })
 
+  it('carries a saved research model over to an agent effort level', () => {
+    const params = ExaBlock.tools.config?.params?.({
+      operation: 'exa_research',
+      model: 'exa-research-pro',
+    }) as Record<string, unknown>
+    expect(params.effort).toBe('high')
+  })
+
+  it('keeps the legacy model sub-block so the serializer preserves its value', () => {
+    const model = ExaBlock.subBlocks.find((block) => block.id === 'model')
+    expect(model?.condition?.value).toBe('exa_research')
+  })
+
+  it('does not map a model value on non-research operations', () => {
+    const params = ExaBlock.tools.config?.params?.({
+      operation: 'exa_agent',
+      model: 'exa-research-pro',
+    }) as Record<string, unknown>
+    expect(params.effort).toBeUndefined()
+  })
+
   it('coerces maxAgeHours of 0 rather than dropping it as falsy', () => {
     const params = ExaBlock.tools.config?.params?.({
       operation: 'exa_search',
@@ -227,6 +248,20 @@ describe('exa_agent terminal statuses', () => {
     expect(result?.error).toMatch(/cancelled/)
   })
 
+  it('emits the retired research output shape so saved references still resolve', async () => {
+    const result = await agentTool.postProcess?.(
+      {
+        success: true,
+        output: { runId: 'agent_run_1', status: 'completed', text: 'the answer' },
+      } as never,
+      { apiKey: API_KEY, query: 'q' } as never,
+      {} as never
+    )
+    expect(result?.output.research).toEqual([
+      { title: 'Research Complete', url: '', summary: 'the answer', text: 'the answer', score: 1 },
+    ])
+  })
+
   it('falls back to the structured payload when a completed run has no text', async () => {
     const result = await settle('completed')
     expect(result?.success).toBe(true)