@outputai/cli 0.12.0 → 0.12.1-next.1243f78.0

package/dist/api/generated/api.js

@@ -1,5 +1,5 @@
 /**
- * Generated by orval v8.9.0 🍺
+ * Generated by orval 🍺
  * Do not edit manually.
  * Output.ai API
  * API for managing and executing Output.ai workflows
@@ -49,6 +49,9 @@ export const WorkflowResultV2ResponseV = {
 export const getGetHealthUrl = () => {
     return `/health`;
 };
+/**
+ * @summary Health check the API
+ */
 export const getHealth = async (options) => {
     return customFetchInstance(getGetHealthUrl(), {
         ...options,
@@ -58,6 +61,9 @@ export const getHealth = async (options) => {
 export const getGetReadyUrl = () => {
     return `/ready`;
 };
+/**
+ * @summary Check if the API is ready to answer requests
+ */
 export const getReady = async (options) => {
     return customFetchInstance(getGetReadyUrl(), {
         ...options,
@@ -67,6 +73,10 @@ export const getReady = async (options) => {
 export const getPostWorkflowRunUrl = () => {
     return `/workflow/run`;
 };
+/**
+ * Executes a workflow and waits for it to complete before returning the result
+ * @summary Execute a workflow synchronously
+ */
 export const postWorkflowRun = async (postWorkflowRunBody, options) => {
     return customFetchInstance(getPostWorkflowRunUrl(), {
         ...options,
@@ -78,6 +88,9 @@ export const postWorkflowRun = async (postWorkflowRunBody, options) => {
 export const getPostWorkflowStartUrl = () => {
     return `/workflow/start`;
 };
+/**
+ * @summary Start a workflow asynchronously
+ */
 export const postWorkflowStart = async (postWorkflowStartBody, options) => {
     return customFetchInstance(getPostWorkflowStartUrl(), {
         ...options,
@@ -89,6 +102,10 @@ export const postWorkflowStart = async (postWorkflowStartBody, options) => {
 export const getGetWorkflowIdStatusUrl = (id) => {
     return `/workflow/${id}/status`;
 };
+/**
+ * Returns the status of the latest run for the given workflow. To pin a specific run, use `/workflow/{id}/runs/{rid}/status`.
+ * @summary Get workflow execution status (latest run)
+ */
 export const getWorkflowIdStatus = async (id, options) => {
     return customFetchInstance(getGetWorkflowIdStatusUrl(id), {
         ...options,
@@ -98,6 +115,9 @@ export const getWorkflowIdStatus = async (id, options) => {
 export const getGetWorkflowIdRunsRidStatusUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/status`;
 };
+/**
+ * @summary Get workflow execution status for a specific run
+ */
 export const getWorkflowIdRunsRidStatus = async (id, rid, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidStatusUrl(id, rid), {
         ...options,
@@ -107,6 +127,9 @@ export const getWorkflowIdRunsRidStatus = async (id, rid, options) => {
 export const getPatchWorkflowIdRunsRidStopUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/stop`;
 };
+/**
+ * @summary Stop a specific workflow run
+ */
 export const patchWorkflowIdRunsRidStop = async (id, rid, options) => {
     return customFetchInstance(getPatchWorkflowIdRunsRidStopUrl(id, rid), {
         ...options,
@@ -116,6 +139,11 @@ export const patchWorkflowIdRunsRidStop = async (id, rid, options) => {
 export const getPatchWorkflowIdStopUrl = (id) => {
     return `/workflow/${id}/stop`;
 };
+/**
+ * Stops the latest run of the given workflow. The returned `runId` reflects the run at describe-time and may differ from the cancelled run if a new execution started concurrently. Deprecated; use `PATCH /workflow/{id}/runs/{rid}/stop` to pin a specific run. Scheduled for removal after 2026-07-16.
+ * @deprecated
+ * @summary [Deprecated] Stop the latest workflow run
+ */
 export const patchWorkflowIdStop = async (id, options) => {
     return customFetchInstance(getPatchWorkflowIdStopUrl(id), {
         ...options,
@@ -125,6 +153,10 @@ export const patchWorkflowIdStop = async (id, options) => {
 export const getPostWorkflowIdRunsRidTerminateUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/terminate`;
 };
+/**
+ * Force terminates a workflow run. Unlike stop/cancel, terminate immediately stops the run without allowing cleanup.
+ * @summary Terminate a specific workflow run (force stop)
+ */
 export const postWorkflowIdRunsRidTerminate = async (id, rid, postWorkflowIdRunsRidTerminateBody, options) => {
     return customFetchInstance(getPostWorkflowIdRunsRidTerminateUrl(id, rid), {
         ...options,
@@ -136,6 +168,11 @@ export const postWorkflowIdRunsRidTerminate = async (id, rid, postWorkflowIdRuns
 export const getPostWorkflowIdTerminateUrl = (id) => {
     return `/workflow/${id}/terminate`;
 };
+/**
+ * Force terminates the latest run. Deprecated; use `POST /workflow/{id}/runs/{rid}/terminate` to target a specific run. Scheduled for removal after 2026-07-16.
+ * @deprecated
+ * @summary [Deprecated] Terminate the latest workflow run
+ */
 export const postWorkflowIdTerminate = async (id, postWorkflowIdTerminateBody, options) => {
     return customFetchInstance(getPostWorkflowIdTerminateUrl(id), {
         ...options,
@@ -147,6 +184,10 @@ export const postWorkflowIdTerminate = async (id, postWorkflowIdTerminateBody, o
 export const getPostWorkflowIdRunsRidResetUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/reset`;
 };
+/**
+ * Resets a pinned workflow run to the point after a completed step, creating a new run that replays from that point. The current execution is terminated.
+ * @summary Reset a specific workflow run to re-run from after a completed step
+ */
 export const postWorkflowIdRunsRidReset = async (id, rid, resetWorkflowRequest, options) => {
     return customFetchInstance(getPostWorkflowIdRunsRidResetUrl(id, rid), {
         ...options,
@@ -158,6 +199,11 @@ export const postWorkflowIdRunsRidReset = async (id, rid, resetWorkflowRequest,
 export const getPostWorkflowIdResetUrl = (id) => {
     return `/workflow/${id}/reset`;
 };
+/**
+ * Resets the latest run. Deprecated; use `POST /workflow/{id}/runs/{rid}/reset` to target a specific run. Scheduled for removal after 2026-07-16.
+ * @deprecated
+ * @summary [Deprecated] Reset the latest workflow run
+ */
 export const postWorkflowIdReset = async (id, resetWorkflowRequest, options) => {
     return customFetchInstance(getPostWorkflowIdResetUrl(id), {
         ...options,
@@ -169,6 +215,10 @@ export const postWorkflowIdReset = async (id, resetWorkflowRequest, options) =>
 export const getGetWorkflowIdResultUrl = (id) => {
     return `/workflow/${id}/result`;
 };
+/**
+ * Returns the result of the latest run for the given workflow. To pin a specific run, use `/workflow/{id}/runs/{rid}/result`.
+ * @summary Return the result of a workflow (latest run)
+ */
 export const getWorkflowIdResult = async (id, options) => {
     return customFetchInstance(getGetWorkflowIdResultUrl(id), {
         ...options,
@@ -178,6 +228,9 @@ export const getWorkflowIdResult = async (id, options) => {
 export const getGetWorkflowIdRunsRidResultUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/result`;
 };
+/**
+ * @summary Return the result of a specific workflow run
+ */
 export const getWorkflowIdRunsRidResult = async (id, rid, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidResultUrl(id, rid), {
         ...options,
@@ -187,6 +240,10 @@ export const getWorkflowIdRunsRidResult = async (id, rid, options) => {
 export const getGetWorkflowIdInputUrl = (id) => {
     return `/workflow/${id}/input`;
 };
+/**
+ * Returns the original input passed to the latest run of the given workflow. Works for workflows in any state, including running. To pin a specific run, use `/workflow/{id}/runs/{rid}/input`.
+ * @summary Return the original input of a workflow (latest run)
+ */
 export const getWorkflowIdInput = async (id, options) => {
     return customFetchInstance(getGetWorkflowIdInputUrl(id), {
         ...options,
@@ -196,6 +253,9 @@ export const getWorkflowIdInput = async (id, options) => {
 export const getGetWorkflowIdRunsRidInputUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/input`;
 };
+/**
+ * @summary Return the original input of a specific workflow run
+ */
 export const getWorkflowIdRunsRidInput = async (id, rid, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidInputUrl(id, rid), {
         ...options,
@@ -205,6 +265,10 @@ export const getWorkflowIdRunsRidInput = async (id, rid, options) => {
 export const getGetWorkflowIdTraceLogUrl = (id) => {
     return `/workflow/${id}/trace-log`;
 };
+/**
+ * Returns trace data for the latest run of the given workflow. If trace is stored remotely (S3), fetches and returns the data inline. If trace is local only, returns the local path. To pin a specific run, use `/workflow/{id}/runs/{rid}/trace-log`.
+ * @summary Get workflow trace log data (latest run)
+ */
 export const getWorkflowIdTraceLog = async (id, options) => {
     return customFetchInstance(getGetWorkflowIdTraceLogUrl(id), {
         ...options,
@@ -214,6 +278,10 @@ export const getWorkflowIdTraceLog = async (id, options) => {
 export const getGetWorkflowIdRunsRidTraceLogUrl = (id, rid) => {
     return `/workflow/${id}/runs/${rid}/trace-log`;
 };
+/**
+ * Returns trace data for a pinned workflow run. If trace is stored remotely (S3), fetches and returns the data inline. If trace is local only, returns the local path.
+ * @summary Get workflow trace log data for a specific run
+ */
 export const getWorkflowIdRunsRidTraceLog = async (id, rid, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidTraceLogUrl(id, rid), {
         ...options,
@@ -224,12 +292,16 @@ export const getGetWorkflowIdHistoryUrl = (id, params) => {
     const normalizedParams = new URLSearchParams();
     Object.entries(params || {}).forEach(([key, value]) => {
         if (value !== undefined) {
-            normalizedParams.append(key, value === null ? 'null' : value.toString());
+            normalizedParams.append(key, value === null ? 'null' : String(value));
         }
     });
     const stringifiedParams = normalizedParams.toString();
     return stringifiedParams.length > 0 ? `/workflow/${id}/history?${stringifiedParams}` : `/workflow/${id}/history`;
 };
+/**
+ * Returns decoded Temporal history events with optional payload inclusion. First page includes workflow metadata; subsequent pages return events only.
+ * @summary Get paginated workflow execution history
+ */
 export const getWorkflowIdHistory = async (id, params, options) => {
     return customFetchInstance(getGetWorkflowIdHistoryUrl(id, params), {
         ...options,
@@ -240,12 +312,15 @@ export const getGetWorkflowIdRunsRidHistoryUrl = (id, rid, params) => {
     const normalizedParams = new URLSearchParams();
     Object.entries(params || {}).forEach(([key, value]) => {
         if (value !== undefined) {
-            normalizedParams.append(key, value === null ? 'null' : value.toString());
+            normalizedParams.append(key, value === null ? 'null' : String(value));
         }
     });
     const stringifiedParams = normalizedParams.toString();
     return stringifiedParams.length > 0 ? `/workflow/${id}/runs/${rid}/history?${stringifiedParams}` : `/workflow/${id}/runs/${rid}/history`;
 };
+/**
+ * @summary Get paginated workflow execution history for a specific run
+ */
 export const getWorkflowIdRunsRidHistory = async (id, rid, params, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidHistoryUrl(id, rid, params), {
         ...options,
@@ -256,12 +331,16 @@ export const getGetWorkflowIdHistoryStreamUrl = (id, params) => {
     const normalizedParams = new URLSearchParams();
     Object.entries(params || {}).forEach(([key, value]) => {
         if (value !== undefined) {
-            normalizedParams.append(key, value === null ? 'null' : value.toString());
+            normalizedParams.append(key, value === null ? 'null' : String(value));
         }
     });
     const stringifiedParams = normalizedParams.toString();
     return stringifiedParams.length > 0 ? `/workflow/${id}/history/stream?${stringifiedParams}` : `/workflow/${id}/history/stream`;
 };
+/**
+ * Opens a persistent SSE connection that delivers Temporal workflow history events in real time. Emits named events: `workflow` (metadata, once), `history` (event batches), `done` (terminal state), `server_error` (post-flush errors). The `done` event carries `{ reason, newRunId? }` where `reason` is the terminal Temporal event type (`WORKFLOW_EXECUTION_COMPLETED`, `WORKFLOW_EXECUTION_FAILED`, `WORKFLOW_EXECUTION_TIMED_OUT`, `WORKFLOW_EXECUTION_CANCELED`, `WORKFLOW_EXECUTION_TERMINATED`, `WORKFLOW_EXECUTION_CONTINUED_AS_NEW`) and `newRunId` is present only when the terminal event chains a follow-on run. `server_error` carries `{ error, message, workflowId, runId }`. Errors before the stream opens are returned as JSON HTTP responses (400/404); once open, failures arrive as a `server_error` event. Supports reconnect via `Last-Event-ID` header or `lastEventId` query param.
+ * @summary Stream workflow history events via Server-Sent Events
+ */
 export const getWorkflowIdHistoryStream = async (id, params, options) => {
     return customFetchInstance(getGetWorkflowIdHistoryStreamUrl(id, params), {
         ...options,
@@ -272,12 +351,16 @@ export const getGetWorkflowIdRunsRidHistoryStreamUrl = (id, rid, params) => {
     const normalizedParams = new URLSearchParams();
     Object.entries(params || {}).forEach(([key, value]) => {
         if (value !== undefined) {
-            normalizedParams.append(key, value === null ? 'null' : value.toString());
+            normalizedParams.append(key, value === null ? 'null' : String(value));
         }
     });
     const stringifiedParams = normalizedParams.toString();
     return stringifiedParams.length > 0 ? `/workflow/${id}/runs/${rid}/history/stream?${stringifiedParams}` : `/workflow/${id}/runs/${rid}/history/stream`;
 };
+/**
+ * Same as /workflow/{id}/history/stream but targets a specific run ID.
+ * @summary Stream pinned-run workflow history events via Server-Sent Events
+ */
 export const getWorkflowIdRunsRidHistoryStream = async (id, rid, params, options) => {
     return customFetchInstance(getGetWorkflowIdRunsRidHistoryStreamUrl(id, rid, params), {
         ...options,
@@ -287,6 +370,9 @@ export const getWorkflowIdRunsRidHistoryStream = async (id, rid, params, options
 export const getGetWorkflowCatalogIdUrl = (id) => {
     return `/workflow/catalog/${id}`;
 };
+/**
+ * @summary Get a specific workflow catalog by ID
+ */
 export const getWorkflowCatalogId = async (id, options) => {
     return customFetchInstance(getGetWorkflowCatalogIdUrl(id), {
         ...options,
@@ -296,6 +382,9 @@ export const getWorkflowCatalogId = async (id, options) => {
 export const getGetWorkflowCatalogUrl = () => {
     return `/workflow/catalog`;
 };
+/**
+ * @summary Get the default workflow catalog
+ */
 export const getWorkflowCatalog = async (options) => {
     return customFetchInstance(getGetWorkflowCatalogUrl(), {
         ...options,
@@ -306,12 +395,16 @@ export const getGetWorkflowRunsUrl = (params) => {
     const normalizedParams = new URLSearchParams();
     Object.entries(params || {}).forEach(([key, value]) => {
         if (value !== undefined) {
-            normalizedParams.append(key, value === null ? 'null' : value.toString());
+            normalizedParams.append(key, value === null ? 'null' : String(value));
         }
     });
     const stringifiedParams = normalizedParams.toString();
     return stringifiedParams.length > 0 ? `/workflow/runs?${stringifiedParams}` : `/workflow/runs`;
 };
+/**
+ * Returns a list of workflow runs with optional filtering by workflow type
+ * @summary List workflow runs
+ */
 export const getWorkflowRuns = async (params, options) => {
     return customFetchInstance(getGetWorkflowRunsUrl(params), {
         ...options,
@@ -321,6 +414,10 @@ export const getWorkflowRuns = async (params, options) => {
 export const getPostWorkflowIdFeedbackUrl = (id) => {
     return `/workflow/${id}/feedback`;
 };
+/**
+ * Always targets the latest run; `runId` cannot be pinned for Temporal signal operations.
+ * @summary Send feedback to a workflow
+ */
 export const postWorkflowIdFeedback = async (id, postWorkflowIdFeedbackBody, options) => {
     return customFetchInstance(getPostWorkflowIdFeedbackUrl(id), {
         ...options,
@@ -332,6 +429,10 @@ export const postWorkflowIdFeedback = async (id, postWorkflowIdFeedbackBody, opt
 export const getPostWorkflowIdSignalSignalUrl = (id, signal) => {
     return `/workflow/${id}/signal/${signal}`;
 };
+/**
+ * Always targets the latest run; `runId` cannot be pinned for Temporal signal operations.
+ * @summary Send a signal to an workflow
+ */
 export const postWorkflowIdSignalSignal = async (id, signal, postWorkflowIdSignalSignalBody, options) => {
     return customFetchInstance(getPostWorkflowIdSignalSignalUrl(id, signal), {
         ...options,
@@ -343,6 +444,10 @@ export const postWorkflowIdSignalSignal = async (id, signal, postWorkflowIdSigna
 export const getPostWorkflowIdQueryQueryUrl = (id, query) => {
     return `/workflow/${id}/query/${query}`;
 };
+/**
+ * Always targets the latest run; `runId` cannot be pinned for Temporal query operations.
+ * @summary Send a query to an workflow
+ */
 export const postWorkflowIdQueryQuery = async (id, query, postWorkflowIdQueryQueryBody, options) => {
     return customFetchInstance(getPostWorkflowIdQueryQueryUrl(id, query), {
         ...options,
@@ -354,6 +459,10 @@ export const postWorkflowIdQueryQuery = async (id, query, postWorkflowIdQueryQue
 export const getPostWorkflowIdUpdateUpdateUrl = (id, update) => {
     return `/workflow/${id}/update/${update}`;
 };
+/**
+ * Always targets the latest run; `runId` cannot be pinned for Temporal update operations.
+ * @summary Execute an update on an workflow
+ */
 export const postWorkflowIdUpdateUpdate = async (id, update, postWorkflowIdUpdateUpdateBody, options) => {
     return customFetchInstance(getPostWorkflowIdUpdateUpdateUrl(id, update), {
         ...options,
@@ -366,6 +475,9 @@ export const postWorkflowIdUpdateUpdate = async (id, update, postWorkflowIdUpdat
 export const getPostHeartbeatUrl = () => {
     return `/heartbeat`;
 };
+/**
+ * @summary A dummy post endpoint for test only
+ */
 export const postHeartbeat = async (options) => {
     return customFetchInstance(getPostHeartbeatUrl(), {
         ...options,

package/dist/assets/docker/docker-compose-dev.yml

@@ -80,7 +80,7 @@ services:
         condition: service_healthy
       worker:
         condition: service_healthy
-    image: outputai/api:${OUTPUT_API_VERSION:-0.12.0}
+    image: outputai/api:${OUTPUT_API_VERSION:-0.12.1-next.1243f78.0}
     init: true
     networks:
       - main

package/dist/generated/framework_version.json

@@ -1,3 +1,3 @@
 {
-    "framework": "0.12.0"
+    "framework": "0.12.1-next.1243f78.0"
 }

package/dist/services/cost_calculator.js

@@ -84,7 +84,8 @@ function parseLegacyLLMUsageEvent(node, stepName, event) {
         model: event.modelId || 'unknown',
         usage: eventTokenUsage(event.usage ?? []),
         originalCost: event.total,
-        lines: event.usage ?? []
+        lines: event.usage ?? [],
+        incomplete: false
     };
 }
 function parseNormalizedLLMUsage(node, stepName, event) {
@@ -100,7 +101,11 @@ function parseNormalizedLLMUsage(node, stepName, event) {
         model: event.modelId || 'unknown',
         usage: normalizedItemTokenUsage(event.items),
         originalCost: 0,
-        lines
+        lines,
+        // This path runs whenever cost computation produced no llm:generation:cost
+        // attribute at all (e.g. pricing config missing, not just unrated grounding),
+        // so every line here is unpriced by construction, not only grounding calls.
+        incomplete: true
     };
 }
 function parseLLMCostEvent(node, stepName, event) {
@@ -116,7 +121,10 @@ function parseLLMCostEvent(node, stepName, event) {
         model: event.modelId || 'unknown',
         usage: normalizedItemTokenUsage(event.items),
         originalCost: event.total ?? 0,
-        lines
+        lines,
+        // An INCOMPLETE event carries a real but unpriced charge (e.g. unrated
+        // grounding). Flag it so the report can mark the understated total.
+        incomplete: event.status === 'incomplete'
     };
 }
 export function extractValue(obj, path) {
@@ -461,7 +469,8 @@ function aggregateLLMCosts(llmCalls, config) {
             cached: call.usage.cachedInputTokens ?? 0,
             reasoning: call.usage.reasoningTokens ?? 0,
             originalCost,
-            adjustedCost
+            adjustedCost,
+            incomplete: call.incomplete
         });
         totals.inputTokens += call.usage.inputTokens ?? 0;
         totals.outputTokens += call.usage.outputTokens ?? 0;

package/dist/services/cost_calculator.spec.js

@@ -37,10 +37,11 @@ function llmEventNode(id, model, lines) {
         }
     };
 }
-function llmCostNode(id, model, items) {
+function llmCostNode(id, model, items, status = 'precise') {
     const totalOf = (group) => items.filter(item => item.group === group).reduce((sum, item) => sum + (item.total ?? 0), 0);
     const input = totalOf('input');
     const output = totalOf('output');
+    const request = totalOf('request');
     return {
         id,
         kind: 'llm',
@@ -52,8 +53,9 @@ function llmCostNode(id, model, items) {
                 modelId: model,
                 input,
                 output,
-                total: input + output,
-                status: 'precise',
+                request,
+                total: input + output + request,
+                status,
                 items
             }
         }
@@ -280,6 +282,66 @@ describe('findLLMCalls', () => {
         ]);
         expect(calls[0].originalCost).toBeCloseTo(0.001965, 8);
     });
+    it('reads a priced grounding request item without counting it as tokens', () => {
+        const items = [
+            { group: 'input', label: 'no_cache', amount: 1032, ppm: 0.3, total: 0.0003096, status: 'ok' },
+            { group: 'output', label: 'text', amount: 793, ppm: 2.5, total: 0.0019825, status: 'ok' },
+            { group: 'request', label: 'grounding_prompt', amount: 1, ppm: 35_000, total: 0.035, status: 'ok' }
+        ];
+        const calls = findLLMCalls({
+            kind: 'workflow',
+            children: [llmCostNode('grounded', 'gemini-2.5-flash', items)]
+        });
+        expect(calls[0].incomplete).toBe(false);
+        expect(calls[0].originalCost).toBeCloseTo(0.0372921, 8);
+        // Grounding bills per request, so it must not inflate token usage.
+        expect(calls[0].usage).toEqual({
+            inputTokens: 1032,
+            cachedInputTokens: 0,
+            outputTokens: 793,
+            reasoningTokens: 0
+        });
+        expect(calls[0].lines.find(l => l.type === 'request_grounding_prompt')).toEqual({
+            type: 'request_grounding_prompt',
+            ppm: 35_000,
+            amount: 1,
+            total: 0.035
+        });
+    });
+    it('preserves a priced grounding charge as-charged when re-pricing at costs.yml rates', () => {
+        const items = [
+            { group: 'input', label: 'no_cache', amount: 1032, ppm: 0.3, total: 0.0003096, status: 'ok' },
+            { group: 'output', label: 'text', amount: 793, ppm: 2.5, total: 0.0019825, status: 'ok' },
+            { group: 'request', label: 'grounding_prompt', amount: 1, ppm: 35_000, total: 0.035, status: 'ok' }
+        ];
+        const config = {
+            models: { 'gemini-2.5-flash': { provider: 'google-vertex', input: 0.3, output: 2.5 } },
+            services: {}
+        };
+        const report = calculateCost({ kind: 'workflow', name: 'w', children: [llmCostNode('grounded', 'gemini-2.5-flash', items)] }, config);
+        // grounding has no costs.yml line rate, so it degrades to its as-charged total, not $0.
+        expect(report.llmAdjustedCost).toBeCloseTo(0.0372921, 8);
+    });
+    it('flags an unrated grounding charge as incomplete instead of a silent $0', () => {
+        const items = [
+            { group: 'input', label: 'no_cache', amount: 1000, ppm: 1, total: 0.001, status: 'ok' },
+            { group: 'output', label: null, amount: 500, ppm: 5, total: 0.0025, status: 'ok' },
+            { group: 'request', label: 'grounding', amount: 3, ppm: null, total: null, status: 'missing' }
+        ];
+        const calls = findLLMCalls({
+            kind: 'workflow',
+            children: [llmCostNode('unrated', 'gemini-9-ultra', items, 'incomplete')]
+        });
+        expect(calls[0].incomplete).toBe(true);
+        // The unrated line still shows up (as $0), but the call carries the signal.
+        expect(calls[0].lines.find(l => l.type === 'request_grounding')).toEqual({
+            type: 'request_grounding',
+            ppm: 0,
+            amount: 3,
+            total: 0
+        });
+        expect(calls[0].usage).toMatchObject({ inputTokens: 1000, outputTokens: 500 });
+    });
     it('reads normalized usage without interpreting token totals as cost', () => {
         const items = [
             { group: 'input', label: null, amount: 100 },
@@ -303,7 +365,9 @@ describe('findLLMCalls', () => {
                 outputTokens: 110,
                 reasoningTokens: 70
             },
-            originalCost: 0
+            originalCost: 0,
+            // Usage-only means cost computation produced no priced result at all.
+            incomplete: true
         });
         expect(calls[0].lines).toEqual([
             { type: 'input', ppm: 0, amount: 100, total: 0 },

package/dist/types/cost.d.ts

@@ -51,6 +51,7 @@ export interface LLMCall {
     usage: TokenUsage;
     originalCost: number;
     lines: LLMUsageLine[];
+    incomplete: boolean;
 }
 export interface HTTPCall {
     stepName: string;
@@ -109,6 +110,7 @@ export interface LLMCostResult {
     reasoning: number;
     originalCost: number;
     adjustedCost: number;
+    incomplete: boolean;
 }
 export interface ServiceCostResult {
     step: string;
@@ -155,6 +157,7 @@ export interface LLMModelSummary {
     count: number;
     originalCost: number;
     adjustedCost: number;
+    incomplete: boolean;
 }
 export interface HostSummary {
     host: string;

package/dist/utils/cost_formatter.js

@@ -25,15 +25,23 @@ function formatCurrency(amount) {
 function pluralize(count, singular) {
     return count === 1 ? `1 ${singular}` : `${count} ${singular}s`;
 }
+// Marks a figure whose call had a real but unpriced charge (e.g. unrated
+// grounding), so a $0.00 line reads as "unpriced" rather than a clean zero.
+const INCOMPLETE_MARKER = '*';
+const INCOMPLETE_FOOTNOTE = '* cost incomplete (unpriced charge such as grounding, or missing usage); the total understates the bill';
+function markIncomplete(text, incomplete) {
+    return incomplete ? `${text} ${INCOMPLETE_MARKER}` : text;
+}
 export function parseCostData(report) {
     const byModel = {};
     for (const r of report.llmCalls) {
         if (!byModel[r.model]) {
-            byModel[r.model] = { count: 0, originalCost: 0, adjustedCost: 0 };
+            byModel[r.model] = { count: 0, originalCost: 0, adjustedCost: 0, incomplete: false };
         }
         byModel[r.model].count++;
         byModel[r.model].originalCost += r.originalCost;
         byModel[r.model].adjustedCost += r.adjustedCost;
+        byModel[r.model].incomplete ||= r.incomplete;
     }
     const llmModels = Object.entries(byModel)
         .sort((a, b) => b[1].adjustedCost - a[1].adjustedCost)
@@ -82,12 +90,13 @@ function formatSummary(data) {
             style: { head: ['cyan'] },
             colAligns: ['left', 'right', 'right', 'right']
         });
+        const anyIncomplete = data.llmModels.some(m => m.incomplete);
         for (const m of data.llmModels) {
             table.push([
                 m.model,
                 pluralize(m.count, 'call'),
-                formatCurrency(m.originalCost),
-                formatCurrency(m.adjustedCost)
+                markIncomplete(formatCurrency(m.originalCost), m.incomplete),
+                markIncomplete(formatCurrency(m.adjustedCost), m.incomplete)
             ]);
         }
         table.push([
@@ -98,6 +107,9 @@ function formatSummary(data) {
         ]);
         lines.push('LLM Costs:');
         lines.push(table.toString());
+        if (anyIncomplete) {
+            lines.push(INCOMPLETE_FOOTNOTE);
+        }
         lines.push('');
     }
     if (data.hosts.length > 0) {
@@ -146,6 +158,7 @@ function formatVerbose(data) {
             style: { head: ['cyan'] },
             colAligns
         });
+        const anyIncomplete = data.llmCalls.some(r => r.incomplete);
         for (const r of data.llmCalls) {
             const row = [
                 r.step,
@@ -159,7 +172,7 @@ function formatVerbose(data) {
             if (data.verbose.hasReasoning) {
                 row.push(formatNumber(r.reasoning));
             }
-            row.push(formatCurrency(r.originalCost), formatCurrency(r.adjustedCost));
+            row.push(markIncomplete(formatCurrency(r.originalCost), r.incomplete), markIncomplete(formatCurrency(r.adjustedCost), r.incomplete));
             table.push(row);
         }
         const totalRow = [
@@ -178,6 +191,9 @@ function formatVerbose(data) {
         table.push(totalRow);
         lines.push('LLM Calls:');
         lines.push(table.toString());
+        if (anyIncomplete) {
+            lines.push(INCOMPLETE_FOOTNOTE);
+        }
         lines.push('');
     }
     if (data.httpDetails.length > 0) {

package/dist/utils/cost_formatter.spec.d.ts

@@ -0,0 +1 @@
+export {};