chore(logs): tighten log-noise changes after a line-by-line audit

- Drop the blockId option threaded through getEffectiveBlockOutputs; it only
  labeled a debug log.
- Name the log-dedupe windows and cache ceilings.
- Build the sandbox catalog route set by converting contract paths to the
  route table's pattern syntax; the route-table test guards drift.

Author: waleedlatif1

apps/sim/executor/utils/block-data.ts

@@ -45,7 +45,6 @@ function getRegistrySchema(block: SerializedBlock): OutputSchema | undefined {
     triggerMode,
     preferToolOutputs: !triggerMode,
     includeHidden: true,
-    blockId: block.id,
   }) as OutputSchema
 
   if (!outputs || Object.keys(outputs).length === 0) {

apps/sim/lib/core/utils/response-format.ts

@@ -78,8 +78,8 @@ export function parseResponseFormatSafely(
     }
     return responseFormatValue
   } catch (error) {
-    // An unparseable response format is user input that falls back to the block's default
-    // outputs; it is re-read on every output-schema derivation, so it is not warning-worthy.
+    // User configuration that falls back to the block's default outputs (the editor's lint
+    // reports it); every output-schema derivation re-reads it, so it is not warning-worthy.
     logger.debug('Failed to parse response format', { blockId, message: getErrorMessage(error) })
     return null
   }

apps/sim/lib/mothership/tools/sandbox-catalog-routes.ts

@@ -5,29 +5,23 @@ import {
   v2ListConnectorTypesContract,
   v2ListToolsContract,
 } from '@/lib/api/contracts/v2/catalog'
-import { matchV2Route } from '@/lib/api/server/routes/in-process-transport'
 
-let catalogRoutePatterns: ReadonlySet<string> | undefined
+/** Catalog contract paths in the generated route table's `{param}` pattern syntax. */
+const CATALOG_ROUTE_PATTERNS: ReadonlySet<string> = new Set(
+  [
+    v2ListBlocksContract,
+    v2GetBlockContract,
+    v2ListToolsContract,
+    v2GetToolContract,
+    v2ListConnectorTypesContract,
+  ].map((contract) => contract.path.replace(/\[([^\]]+)\]/g, '{$1}'))
+)
 
 /**
  * Whether a sandbox API request reads the producer-owned catalog (blocks, tools, connector
  * types): no execution output, file content, or row values, so no producer records secret
- * provenance for it and none is expected. Each contract resolves through the generated route
- * table, so a renamed path parameter cannot drift from the pattern `matchV2Route` reports.
+ * provenance for it and none is expected.
  */
 export function isCatalogRoute(method: string, routePattern: string): boolean {
-  if (method !== 'GET') return false
-  catalogRoutePatterns ??= new Set(
-    [
-      v2ListBlocksContract,
-      v2GetBlockContract,
-      v2ListToolsContract,
-      v2GetToolContract,
-      v2ListConnectorTypesContract,
-    ].flatMap((contract) => {
-      const pattern = matchV2Route(contract.path.replace(/\[[^\]]+\]/g, 'catalog-id'))?.pattern
-      return pattern ? [pattern] : []
-    })
-  )
-  return catalogRoutePatterns.has(routePattern)
+  return method === 'GET' && CATALOG_ROUTE_PATTERNS.has(routePattern)
 }

apps/sim/lib/mothership/tools/sandbox-resource-transport.ts

@@ -30,13 +30,18 @@ import { observeWorkspaceFileDelivery } from '@/lib/workspace-files/application/
 
 const logger = createLogger('MothershipSandboxResourceTransport')
 
+/** How long one route's missing-provenance report suppresses repeats for that route. */
+const UNRECORDED_PROVENANCE_LOG_WINDOW_MS = 60 * 60 * 1000
+/** Memory backstop above the number of `method route` pairs the v2 route table can produce. */
+const UNRECORDED_PROVENANCE_LOG_MAX_ROUTES = 1_000
+
 /**
  * Data-bearing routes already reported as lacking a provenance producer. The gap is a
  * property of the route, not the request, so it is reported once per route per window.
  */
 const unrecordedProvenanceRoutesLogged = new LRUCache<string, true>({
-  max: 1_000,
-  ttl: 60 * 60 * 1000,
+  max: UNRECORDED_PROVENANCE_LOG_MAX_ROUTES,
+  ttl: UNRECORDED_PROVENANCE_LOG_WINDOW_MS,
 })
 
 /** Private callback observes the real authenticated v2 request without changing its body or API contract. */

apps/sim/lib/tokenization/accurate.ts

@@ -25,11 +25,16 @@ const logger = createLogger('TokenizationAccurate')
 /** Keyed by encoding, not model: each instance holds a full rank table, so models share one. */
 const encodingCache = new Map<TiktokenEncoding, Tiktoken>()
 
+/** Memory backstop for {@link encodingNameByModel}; model ids are caller-supplied strings. */
+const ENCODING_NAME_CACHE_MAX_MODELS = 1_000
+
 /**
  * Model id → encoding name, so a non-OpenAI model does not throw and catch inside
- * `getEncodingNameForModel` on every count. Model ids are caller-supplied, hence the ceiling.
+ * `getEncodingNameForModel` on every count.
  */
-const encodingNameByModel = new LRUCache<string, TiktokenEncoding>({ max: 1_000 })
+const encodingNameByModel = new LRUCache<string, TiktokenEncoding>({
+  max: ENCODING_NAME_CACHE_MAX_MODELS,
+})
 
 /** OpenAI families tokenized with `o200k_base` that `js-tiktoken`'s exact-name table may not list yet. */
 const O200K_MODEL_FAMILY = /^(?:gpt-(?:4o|4\.1|4\.5|5|6|oss)|chatgpt-4o|o\d)/

apps/sim/lib/webhooks/providers/slack.ts

@@ -408,15 +408,20 @@ async function downloadSlackFiles(
   return downloaded
 }
 
+/** How long a webhook's `missing_scope` report suppresses repeats of the same log line. */
+const MISSING_SCOPE_LOG_WINDOW_MS = 60 * 60 * 1000
+/** Memory backstop for {@link reactionsMissingScopeLogged}, far above live Slack webhooks. */
+const MISSING_SCOPE_LOG_MAX_WEBHOOKS = 10_000
+
 /**
  * Webhooks whose bot already reported `missing_scope` on `reactions.get`. A bot without
  * `reactions:read` fails every reaction event identically until it is reinstalled, so the
  * configuration problem is logged once per webhook per window rather than per event. The call
  * itself still runs, so a reinstalled bot gets message text back immediately.
  */
 const reactionsMissingScopeLogged = new LRUCache<string, true>({
-  max: 10_000,
-  ttl: 60 * 60 * 1000,
+  max: MISSING_SCOPE_LOG_MAX_WEBHOOKS,
+  ttl: MISSING_SCOPE_LOG_WINDOW_MS,
 })
 
 async function fetchSlackMessageText(

apps/sim/lib/workflows/blocks/block-outputs.ts

@@ -38,8 +38,6 @@ interface EffectiveOutputOptions {
   triggerMode?: boolean
   preferToolOutputs?: boolean
   includeHidden?: boolean
-  /** The block instance id, used only to attribute an unparseable response format. */
-  blockId?: string
 }
 
 type ConditionValue = string | number | boolean
@@ -434,7 +432,6 @@ export function getEffectiveBlockOutputs(
   const includeHidden = options?.includeHidden ?? false
 
   if (blockType === 'agent') {
-    const responseFormatBlockId = options?.blockId ?? blockType
     const model = subBlocks?.model?.value
     const mayEvaluate =
       typeof model === 'string' && (isEvaluationModel(model) || containsReference(model))
@@ -443,13 +440,11 @@ export function getEffectiveBlockOutputs(
       const answers = getJevAnswerOutput(subBlocks?.evaluationQuestions?.value)
       return {
         ...outputs,
-        ...(containsReference(model)
-          ? getResponseFormatOutputs(subBlocks, responseFormatBlockId)
-          : undefined),
+        ...(containsReference(model) ? getResponseFormatOutputs(subBlocks, 'agent') : undefined),
         ...(answers ? { answers } : undefined),
       }
     }
-    const responseFormatOutputs = getResponseFormatOutputs(subBlocks, responseFormatBlockId)
+    const responseFormatOutputs = getResponseFormatOutputs(subBlocks, 'agent')
     if (responseFormatOutputs) return responseFormatOutputs
   }