@outputai/cli 0.7.1-next.ae5bab4.0 → 0.7.1-next.bd6bd49.0

package/dist/services/cost_calculator.spec.js

@@ -9,7 +9,46 @@ const mockLoad = vi.fn();
 vi.mock('js-yaml', () => ({
     default: { load: (...args) => mockLoad(...args) }
 }));
-import { extractValue, findLLMCalls, findHTTPCalls, calculateLLMCallCost, identifyService, calculateServiceCost, calculateCost, loadPricingConfig } from '#services/cost_calculator.js';
+import { extractValue, findLLMCalls, findHTTPCalls, identifyService, calculateServiceCost, calculateCost, loadPricingConfig } from '#services/cost_calculator.js';
+// ---------------------------------------------------------------------------
+// Fixture builders
+// ---------------------------------------------------------------------------
+function llmLines(inAmt, inPpm, outAmt, outPpm, cachedAmt = 0, cachedPpm = 0) {
+    return [
+        { type: 'input', ppm: inPpm, amount: inAmt, total: (inAmt / 1e6) * inPpm },
+        { type: 'input_cached', ppm: cachedPpm, amount: cachedAmt, total: (cachedAmt / 1e6) * cachedPpm },
+        { type: 'output', ppm: outPpm, amount: outAmt, total: (outAmt / 1e6) * outPpm }
+    ];
+}
+function llmEventNode(id, model, lines) {
+    const total = lines.reduce((s, l) => s + l.total, 0);
+    return {
+        id,
+        kind: 'llm',
+        name: 'gen',
+        attributes: {
+            'llm:usage': {
+                type: 'llm:usage',
+                modelId: model,
+                usage: lines,
+                total,
+                tokensUsed: lines.reduce((s, l) => s + l.amount, 0)
+            }
+        }
+    };
+}
+function httpEventNode(id, url, method, total, output = { status: 200, body: {} }, inputBody) {
+    return {
+        id,
+        kind: 'http',
+        name: 'request',
+        input: { url, method, ...(inputBody ? { body: inputBody } : {}) },
+        output,
+        attributes: {
+            'http:request:cost': { type: 'http:request:cost', url, requestId: id, total }
+        }
+    };
+}
 const llmTrace = {
     id: 'test-trace-1',
     kind: 'workflow',
@@ -21,29 +60,13 @@ const llmTrace = {
             id: 'step-1',
             kind: 'step',
             name: 'test_workflow#generate_summary',
-            children: [
-                {
-                    id: 'llm-1',
-                    kind: 'llm',
-                    name: 'generate_summary',
-                    input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5' } } },
-                    output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                }
-            ]
+            children: [llmEventNode('llm-1', 'claude-sonnet-4-5', llmLines(1000, 3, 500, 15))]
         },
         {
             id: 'step-2',
             kind: 'step',
             name: 'test_workflow#analyze_data',
-            children: [
-                {
-                    id: 'llm-2',
-                    kind: 'llm',
-                    name: 'analyze_data',
-                    input: { loadedPrompt: { config: { model: 'claude-haiku-4-5' } } },
-                    output: { usage: { inputTokens: 2000, outputTokens: 1000, cachedInputTokens: 500 } }
-                }
-            ]
+            children: [llmEventNode('llm-2', 'claude-haiku-4-5', llmLines(1500, 1, 1000, 5, 500, 0.1))]
         }
     ]
 };
@@ -58,35 +81,13 @@ const httpTrace = {
             id: 'step-1',
             kind: 'step',
             name: 'test_workflow#fetch_content',
-            children: [
-                {
-                    id: 'http-1',
-                    kind: 'http',
-                    name: 'jina_request',
-                    input: { url: 'https://r.jina.ai/https://example.com', method: 'GET' },
-                    output: { status: 200, body: { data: { usage: { tokens: 5000 } } } }
-                }
-            ]
+            children: [httpEventNode('http-1', 'https://r.jina.ai/https://example.com', 'GET', 0.0002)]
         },
         {
             id: 'step-2',
             kind: 'step',
             name: 'test_workflow#search',
-            children: [
-                {
-                    id: 'http-2',
-                    kind: 'http',
-                    name: 'exa_request',
-                    input: { url: 'https://api.exa.ai/research', method: 'GET' },
-                    output: {
-                        status: 200,
-                        body: {
-                            model: 'exa-research',
-                            costDollars: { total: 0.15, numSearches: 1, numPages: 5 }
-                        }
-                    }
-                }
-            ]
+            children: [httpEventNode('http-2', 'https://api.exa.ai/research', 'POST', 0.012)]
         }
     ]
 };
@@ -101,36 +102,21 @@ const duplicateTrace = {
             id: 'step-1',
             kind: 'step',
             name: 'test_workflow#step_one',
-            children: [
-                {
-                    id: 'llm-same-id',
-                    kind: 'llm',
-                    name: 'generate',
-                    input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5' } } },
-                    output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                }
-            ]
+            children: [llmEventNode('llm-same-id', 'claude-sonnet-4-5', llmLines(1000, 3, 500, 15))]
         },
         {
             id: 'child-workflow',
             kind: 'workflow',
             name: 'child_workflow',
-            children: [
-                {
-                    id: 'llm-same-id',
-                    kind: 'llm',
-                    name: 'generate',
-                    input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5' } } },
-                    output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                }
-            ]
+            children: [llmEventNode('llm-same-id', 'claude-sonnet-4-5', llmLines(1000, 3, 500, 15))]
         }
     ]
 };
 const testConfig = {
     models: {
         'claude-sonnet-4-5': { provider: 'anthropic', input: 3.0, output: 15.0, cached_input: 0.30 },
-        'claude-haiku-4-5': { provider: 'anthropic', input: 1.0, output: 5.0, cached_input: 0.10 }
+        'claude-haiku-4-5': { provider: 'anthropic', input: 1.0, output: 5.0, cached_input: 0.10 },
+        'claude-opus-4': { provider: 'anthropic', input: 15.0, output: 75.0, cached_input: 1.50 }
     },
     services: {
         jina: {
@@ -144,6 +130,14 @@ const testConfig = {
             url_pattern: 'api.exa.ai',
             cost_path: 'output.body.costDollars.total',
             billable_method: 'POST'
+        },
+        tavily: {
+            type: 'request',
+            url_pattern: 'api.tavily.com',
+            endpoints: {
+                search: { pattern: '/search', price: 0.01 },
+                extract: { pattern: '/extract', price_per_item: 0.005, items_path: 'body.urls' }
+            }
         }
     }
 };
@@ -167,29 +161,43 @@ describe('extractValue', () => {
     });
 });
 describe('findLLMCalls', () => {
-    it('finds nested LLM calls', () => {
+    it('finds nested LLM calls and extracts step names', () => {
         const calls = findLLMCalls(llmTrace);
         expect(calls).toHaveLength(2);
-    });
-    it('extracts step names correctly', () => {
-        const calls = findLLMCalls(llmTrace);
         expect(calls[0].stepName).toBe('generate_summary');
         expect(calls[1].stepName).toBe('analyze_data');
-    });
-    it('extracts model names from loadedPrompt config', () => {
-        const calls = findLLMCalls(llmTrace);
         expect(calls[0].model).toBe('claude-sonnet-4-5');
-        expect(calls[1].model).toBe('claude-haiku-4-5');
-    });
-    it('extracts token usage', () => {
-        const calls = findLLMCalls(llmTrace);
-        expect(calls[0].usage.inputTokens).toBe(1000);
-        expect(calls[0].usage.outputTokens).toBe(500);
         expect(calls[1].usage.cachedInputTokens).toBe(500);
     });
-    it('deduplicates by ID', () => {
-        const calls = findLLMCalls(duplicateTrace);
+    it('reads model, tokens and as-charged cost from llm:usage event', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [llmEventNode('llm-ev', 'gpt-5.5', llmLines(27400, 5, 7275, 30))]
+        };
+        const calls = findLLMCalls(trace);
         expect(calls).toHaveLength(1);
+        expect(calls[0].model).toBe('gpt-5.5');
+        expect(calls[0].usage.inputTokens).toBe(27400);
+        expect(calls[0].usage.outputTokens).toBe(7275);
+        expect(calls[0].originalCost).toBeCloseTo(0.35525, 5);
+    });
+    it('ignores llm nodes without a llm:usage event', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [{
+                    id: 'llm-old',
+                    kind: 'llm',
+                    name: 'gen',
+                    input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5' } } },
+                    output: { usage: { inputTokens: 1000, outputTokens: 500 } }
+                }]
+        };
+        expect(findLLMCalls(trace)).toHaveLength(0);
+    });
+    it('deduplicates nodes by ID', () => {
+        expect(findLLMCalls(duplicateTrace)).toHaveLength(1);
     });
 });
 describe('findHTTPCalls', () => {
@@ -197,71 +205,43 @@ describe('findHTTPCalls', () => {
         const calls = findHTTPCalls(httpTrace);
         expect(calls).toHaveLength(2);
     });
-    it('extracts URLs and methods', () => {
+    it('extracts URLs, methods and host', () => {
         const calls = findHTTPCalls(httpTrace);
         expect(calls[0].url).toContain('r.jina.ai');
         expect(calls[0].method).toBe('GET');
+        expect(calls[0].host).toBe('r.jina.ai');
     });
     it('extracts step names', () => {
         const calls = findHTTPCalls(httpTrace);
         expect(calls[0].stepName).toBe('fetch_content');
         expect(calls[1].stepName).toBe('search');
     });
-});
-describe('calculateLLMCallCost', () => {
-    it('calculates cost for known model', () => {
-        const usage = { inputTokens: 1000000, outputTokens: 500000 };
-        const modelPricing = { provider: 'anthropic', input: 3.0, output: 15.0 };
-        const { cost } = calculateLLMCallCost(usage, modelPricing);
-        // 1M input * $3/M + 0.5M output * $15/M = $3 + $7.5 = $10.5
-        expect(cost).toBeCloseTo(10.5, 2);
-    });
-    it('includes cached input tokens at reduced rate', () => {
-        const usage = { inputTokens: 1000000, outputTokens: 0, cachedInputTokens: 1000000 };
-        const modelPricing = { provider: 'anthropic', input: 3.0, output: 15.0, cached_input: 0.3 };
-        const { cost } = calculateLLMCallCost(usage, modelPricing);
-        // 1M input * $3/M + 1M cached * $0.3/M = $3 + $0.3 = $3.3
-        expect(cost).toBeCloseTo(3.3, 2);
-    });
-    it('returns zero with warning for unknown model', () => {
-        const usage = { inputTokens: 1000, outputTokens: 500 };
-        const { cost, warning } = calculateLLMCallCost(usage, undefined);
-        expect(cost).toBe(0);
-        expect(warning).toBe('unknown model');
+    it('reads host and as-charged cost from http:request:cost event', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('req-1', 'https://api.exa.ai/search', 'POST', 0.012)]
+        };
+        const calls = findHTTPCalls(trace);
+        expect(calls).toHaveLength(1);
+        expect(calls[0].host).toBe('api.exa.ai');
+        expect(calls[0].originalCost).toBe(0.012);
     });
 });
 describe('identifyService', () => {
+    const baseCall = (url) => ({
+        stepName: 'test', url, method: 'GET', input: {}, output: {}, host: url
+    });
     it('identifies Jina by URL pattern', () => {
-        const call = {
-            stepName: 'test',
-            url: 'https://r.jina.ai/https://example.com',
-            method: 'GET',
-            input: {},
-            output: {}
-        };
-        const result = identifyService(call, testConfig.services);
+        const result = identifyService(baseCall('https://r.jina.ai/https://example.com'), testConfig.services);
         expect(result?.serviceName).toBe('jina');
     });
     it('identifies Exa by URL pattern', () => {
-        const call = {
-            stepName: 'test',
-            url: 'https://api.exa.ai/research',
-            method: 'GET',
-            input: {},
-            output: {}
-        };
-        const result = identifyService(call, testConfig.services);
+        const result = identifyService(baseCall('https://api.exa.ai/research'), testConfig.services);
         expect(result?.serviceName).toBe('exa');
     });
     it('returns null for unknown URLs', () => {
-        const call = {
-            stepName: 'test',
-            url: 'https://unknown-api.com/endpoint',
-            method: 'GET',
-            input: {},
-            output: {}
-        };
-        const result = identifyService(call, testConfig.services);
+        const result = identifyService(baseCall('https://unknown-api.com/endpoint'), testConfig.services);
         expect(result).toBeNull();
     });
 });
@@ -271,10 +251,9 @@ describe('calculateServiceCost', () => {
             stepName: 'test',
             url: 'https://r.jina.ai/https://example.com',
             method: 'GET',
+            host: 'r.jina.ai',
             input: {},
-            output: {
-                body: { data: { usage: { tokens: 1000000 } } }
-            }
+            output: { body: { data: { usage: { tokens: 1000000 } } } }
         };
         const serviceInfo = identifyService(call, testConfig.services);
         const result = calculateServiceCost(call, serviceInfo);
@@ -286,12 +265,10 @@ describe('calculateServiceCost', () => {
             stepName: 'test',
             url: 'https://api.exa.ai/research',
             method: 'GET',
+            host: 'api.exa.ai',
             input: {},
             output: {
-                body: {
-                    model: 'exa-research',
-                    costDollars: { total: 0.15, numSearches: 1, numPages: 5 }
-                }
+                body: { model: 'exa-research', costDollars: { total: 0.15, numSearches: 1, numPages: 5 } }
             }
         };
         const serviceInfo = identifyService(call, testConfig.services);
@@ -305,10 +282,9 @@ describe('calculateServiceCost', () => {
             stepName: 'test',
             url: 'https://api.exa.ai/research',
             method: 'GET',
+            host: 'api.exa.ai',
             input: {},
-            output: {
-                body: { status: 'pending' }
-            }
+            output: { body: { status: 'pending' } }
         };
         const serviceInfo = identifyService(call, testConfig.services);
         const result = calculateServiceCost(call, serviceInfo);
@@ -316,92 +292,22 @@ describe('calculateServiceCost', () => {
         expect(result.warning).toBe('no cost data');
     });
 });
-describe('response_cost filtering in calculateCost', () => {
-    it('skips Exa polling requests without cost data', () => {
-        const trace = {
-            kind: 'workflow',
-            name: 'test_workflow',
-            startedAt: 1700000000000,
-            endedAt: 1700000100000,
-            children: [
-                {
-                    id: 'step-exa',
-                    kind: 'step',
-                    name: 'test_workflow#search',
-                    children: [
-                        {
-                            id: 'http-exa-poll',
-                            kind: 'http',
-                            name: 'exa_poll',
-                            input: { url: 'https://api.exa.ai/research/task-123', method: 'GET' },
-                            output: { status: 200, body: { status: 'in_progress' } }
-                        }
-                    ]
-                }
-            ]
-        };
-        const report = calculateCost(trace, testConfig, 'test.json');
-        expect(report.serviceTotalCost).toBe(0);
-        expect(report.services).toHaveLength(0);
-    });
-    it('counts Exa responses that have costDollars', () => {
-        const trace = {
-            kind: 'workflow',
-            name: 'test_workflow',
-            startedAt: 1700000000000,
-            endedAt: 1700000100000,
-            children: [
-                {
-                    id: 'step-exa',
-                    kind: 'step',
-                    name: 'test_workflow#search',
-                    children: [
-                        {
-                            id: 'http-exa-poll',
-                            kind: 'http',
-                            name: 'exa_poll',
-                            input: { url: 'https://api.exa.ai/research/task-123', method: 'GET' },
-                            output: { status: 200, body: { status: 'in_progress' } }
-                        },
-                        {
-                            id: 'http-exa-result',
-                            kind: 'http',
-                            name: 'exa_result',
-                            input: { url: 'https://api.exa.ai/research/task-123', method: 'GET' },
-                            output: {
-                                status: 200,
-                                body: {
-                                    model: 'exa-research',
-                                    costDollars: { total: 0.08, numSearches: 1, numPages: 3 }
-                                }
-                            }
-                        }
-                    ]
-                }
-            ]
-        };
-        const report = calculateCost(trace, testConfig, 'test.json');
-        expect(report.services).toHaveLength(1);
-        expect(report.services[0].calls).toHaveLength(1);
-        expect(report.services[0].totalCost).toBeCloseTo(0.08, 4);
-    });
-});
 describe('calculateCost', () => {
     it('calculates total cost for LLM trace', () => {
         const report = calculateCost(llmTrace, testConfig, 'test.json');
         expect(report.llmCalls).toHaveLength(2);
         expect(report.workflowName).toBe('test_workflow');
-        expect(report.llmTotalCost).toBeGreaterThan(0);
-        expect(report.totalCost).toBe(report.llmTotalCost + report.serviceTotalCost);
+        expect(report.llmAdjustedCost).toBeGreaterThan(0);
+        expect(report.totalCost).toBe(report.llmAdjustedCost + report.httpAdjustedCost);
+        expect(report.originalTotalCost).toBe(report.llmOriginalCost + report.httpOriginalCost);
     });
     it('calculates total cost for HTTP trace', () => {
         const report = calculateCost(httpTrace, testConfig, 'test.json');
-        expect(report.services.length).toBeGreaterThan(0);
-        expect(report.serviceTotalCost).toBeGreaterThan(0);
+        expect(report.httpCosts.length).toBeGreaterThan(0);
+        expect(report.httpAdjustedCost).toBeGreaterThan(0);
     });
     it('calculates duration from timestamps', () => {
         const report = calculateCost(llmTrace, testConfig, 'test.json');
-        // 1700000100000 - 1700000000000 = 100000ms
         expect(report.durationMs).toBe(100000);
     });
     it('handles deduplication in cost calculation', () => {
@@ -412,65 +318,235 @@ describe('calculateCost', () => {
         const trace = {
             kind: 'workflow',
             name: 'test',
-            children: [{
-                    id: 'step-1',
-                    kind: 'step',
-                    name: 'test#gen',
-                    children: [{
-                            id: 'llm-1',
-                            kind: 'llm',
-                            name: 'gen',
-                            input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5-20250514' } } },
-                            output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                        }]
-                }]
+            children: [
+                llmEventNode('llm-1', 'claude-sonnet-4-5-20250514', llmLines(1000, 3, 500, 15))
+            ]
         };
         const report = calculateCost(trace, testConfig, 'test.json');
-        expect(report.llmTotalCost).toBeGreaterThan(0);
-        expect(report.unknownModels).toHaveLength(0);
-        expect(report.llmCalls[0].warning).toBe('priced as claude-sonnet-4-5');
+        // priced at the claude-sonnet-4-5 prefix rates: 1000@$3/M + 500@$15/M
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(0.0105, 5);
     });
-    it('reports unknown model when no prefix match exists', () => {
+});
+describe('event-driven LLM costs (original vs adjusted)', () => {
+    it('matches original when costs.yml rate equals the event rate', () => {
+        // haiku event priced at the same rate as testConfig (input 1 / output 5)
         const trace = {
             kind: 'workflow',
             name: 'test',
-            children: [{
-                    id: 'step-1',
-                    kind: 'step',
-                    name: 'test#gen',
-                    children: [{
-                            id: 'llm-1',
-                            kind: 'llm',
-                            name: 'gen',
-                            input: { loadedPrompt: { config: { model: 'totally-unknown-model' } } },
-                            output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                        }]
-                }]
+            children: [llmEventNode('h', 'claude-haiku-4-5', llmLines(1_000_000, 1, 1_000_000, 5))]
         };
         const report = calculateCost(trace, testConfig, 'test.json');
-        expect(report.llmTotalCost).toBe(0);
-        expect(report.unknownModels).toContain('totally-unknown-model');
-        expect(report.llmCalls[0].warning).toBe('unknown model');
+        expect(report.llmCalls[0].originalCost).toBeCloseTo(6, 5);
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(6, 5);
     });
-    it('prefers exact model match over prefix', () => {
+    it('overrides via prefix match when the configured rate differs (opus-4-8 → opus-4)', () => {
+        // event charged at 5/25; costs.yml has no opus-4-8, prefix-matches opus-4 at 15/75
         const trace = {
             kind: 'workflow',
             name: 'test',
-            children: [{
-                    id: 'step-1',
-                    kind: 'step',
-                    name: 'test#gen',
-                    children: [{
-                            id: 'llm-1',
-                            kind: 'llm',
-                            name: 'gen',
-                            input: { loadedPrompt: { config: { model: 'claude-sonnet-4-5' } } },
-                            output: { usage: { inputTokens: 1000, outputTokens: 500 } }
-                        }]
-                }]
+            children: [llmEventNode('o', 'claude-opus-4-8', llmLines(1_000_000, 5, 1_000_000, 25))]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.llmCalls[0].originalCost).toBeCloseTo(30, 5);
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(90, 5);
+        expect(report.llmOriginalCost).toBeCloseTo(30, 5);
+        expect(report.llmAdjustedCost).toBeCloseTo(90, 5);
+    });
+    it('leaves adjusted equal to original for a model not in costs.yml (gpt-5.5)', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [llmEventNode('g', 'gpt-5.5', llmLines(27400, 5, 7275, 30))]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.llmCalls[0].originalCost).toBeCloseTo(0.35525, 5);
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(0.35525, 5);
+    });
+    it('prices an unknown line type at its as-charged total, not $0', () => {
+        const lines = [
+            ...llmLines(1_000_000, 1, 1_000_000, 5),
+            { type: 'input_cache_write', ppm: 1.25, amount: 200_000, total: 0.25 }
+        ];
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [llmEventNode('cw', 'claude-haiku-4-5', lines)]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        // known lines reprice to $6 at config rates; unknown line passes through at $0.25
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(6.25, 5);
+        expect(report.llmCalls[0].originalCost).toBeCloseTo(6.25, 5);
+    });
+    it('reports inputTokens including cached tokens for event traces', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [llmEventNode('c', 'claude-haiku-4-5', llmLines(1000, 1, 500, 5, 600, 0.1))]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        // producer 'input' line excludes cached — display restores the AI SDK total
+        expect(report.llmCalls[0].input).toBe(1600);
+        expect(report.llmCalls[0].cached).toBe(600);
+        // repriced: 1000@$1/M + 600@$0.1/M + 500@$5/M
+        expect(report.llmCalls[0].adjustedCost).toBeCloseTo(0.00356, 6);
+    });
+});
+describe('event-driven HTTP costs (original vs adjusted)', () => {
+    it('falls back to as-charged cost when service recompute is not usable', () => {
+        // exa POST with a cost event but no costDollars body → recompute is $0 → use event
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('exa-1', 'https://api.exa.ai/search', 'POST', 0.012)]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts).toHaveLength(1);
+        expect(report.httpCosts[0].host).toBe('api.exa.ai');
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.012, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.012, 5);
+    });
+    it('overrides with the recomputed cost for a request-based service (tavily)', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('tv-1', 'https://api.tavily.com/search', 'POST', 0.008)]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts[0].host).toBe('api.tavily.com');
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.008, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.01, 5);
+    });
+    it('uses as-charged cost for an unconfigured host (firecrawl)', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('fc-1', 'https://api.firecrawl.dev/v1/scrape', 'POST', 0.0008)]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts[0].host).toBe('api.firecrawl.dev');
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.0008, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.0008, 5);
+    });
+    it('counts an event-bearing 4xx call as-charged (the event proves a charge)', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('fc-429', 'https://api.firecrawl.dev/v1/scrape', 'POST', 0.10, { status: 429, body: {} })]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts).toHaveLength(1);
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.10, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.10, 5);
+    });
+    it('excludes eventless calls — only http:request:cost events are billable', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [
+                httpEventNode('fc-1', 'https://api.firecrawl.dev/v1/scrape', 'POST', 0.05),
+                // configured service with priceable body, but no cost event → not billed
+                {
+                    id: 'jina-no-event',
+                    kind: 'http',
+                    name: 'request',
+                    input: { url: 'https://r.jina.ai/https://example.com', method: 'GET' },
+                    output: { status: 200, body: { data: { usage: { tokens: 1000000 } } } }
+                }
+            ]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts.find(h => h.host === 'r.jina.ai')).toBeUndefined();
+        expect(report.httpCosts.find(h => h.host === 'api.firecrawl.dev')).toBeDefined();
+    });
+    it('applies a legitimately computed $0 override', () => {
+        const config = {
+            ...testConfig,
+            services: {
+                ...testConfig.services,
+                tavily: {
+                    type: 'request',
+                    url_pattern: 'api.tavily.com',
+                    endpoints: { search: { pattern: '/search', price: 0 } }
+                }
+            }
+        };
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('tv-free', 'https://api.tavily.com/search', 'POST', 0.008)]
+        };
+        const report = calculateCost(trace, config, 'test.json');
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.008, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBe(0);
+    });
+    it('does not let a fallback estimate override an exact event cost', () => {
+        const config = {
+            ...testConfig,
+            services: {
+                ...testConfig.services,
+                exa: {
+                    ...testConfig.services.exa,
+                    fallback_models: { 'exa-research': 0.10 },
+                    default_fallback: 0.10
+                }
+            }
+        };
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('exa-est', 'https://api.exa.ai/search', 'POST', 0.012)]
+        };
+        const report = calculateCost(trace, config, 'test.json');
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.012, 5);
+    });
+    it('treats an un-captured request body as failed, preserving the as-charged cost', () => {
+        // tavily /extract is price_per_item over body.urls — body missing from trace
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('tv-x', 'https://api.tavily.com/extract', 'POST', 0.015)]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts[0].adjustedTotalCost).toBeCloseTo(0.015, 5);
+    });
+    it('overrides with a real measured item count, including an empty array', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [httpEventNode('tv-0', 'https://api.tavily.com/extract', 'POST', 0.015, { status: 200, body: {} }, { urls: [] })]
+        };
+        const report = calculateCost(trace, testConfig, 'test.json');
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.015, 5);
+        expect(report.httpCosts[0].adjustedTotalCost).toBe(0);
+    });
+    it('ignores count-only nodes and groups billable requests by host', () => {
+        const trace = {
+            kind: 'workflow',
+            name: 'test',
+            children: [
+                // count-only webhook (no cost event) — must be excluded
+                {
+                    id: 'wh-1',
+                    kind: 'http',
+                    name: 'request',
+                    input: { url: 'https://os.growthx.ai/webhooks/output', method: 'POST' },
+                    attributes: {
+                        'http:request:count': {
+                            type: 'http:request:count',
+                            url: 'https://os.growthx.ai/webhooks/output',
+                            requestId: 'wh-1'
+                        }
+                    }
+                },
+                httpEventNode('fc-1', 'https://api.firecrawl.dev/v1/scrape', 'POST', 0.10),
+                httpEventNode('fc-2', 'https://api.firecrawl.dev/v1/scrape', 'POST', 0.05)
+            ]
         };
         const report = calculateCost(trace, testConfig, 'test.json');
-        expect(report.llmCalls[0].warning).toBeUndefined();
+        expect(report.httpCosts).toHaveLength(1);
+        expect(report.httpCosts[0].host).toBe('api.firecrawl.dev');
+        expect(report.httpCosts[0].calls).toHaveLength(2);
+        expect(report.httpCosts[0].originalTotalCost).toBeCloseTo(0.15, 5);
     });
 });
 describe('loadPricingConfig', () => {