{ "schemaVersion": 1, "reportTitle": "Benchmark Results", "reportSubtitle": "Multi-task comparison: js-project-copperline-find-vulns, js-project-goldleaf-find-vulns, js-project-ironclad-find-vulns, js-project-nightowl-find-vulns, js-project-purplehaze-find-vulns, js-project-riverbend-find-vulns, js-project-shadowfox-find-vulns, js-project-silvergate-find-vulns, js-project-skylark-find-vulns, js-project-tigerteam-find-vulns", "generatedAt": "2026-05-25T09:12:47.212Z", "sourceFiles": [ "results/benchmark-2026-05-20T23-06-29-348Z.jsonl" ], "htmlReport": "index.html", "charts": [ { "id": "headline-score", "title": "Headline score", "chartType": "bar", "scope": "config-aggregate", "metric": "score", "unit": "percent", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-score", "caption": "Macro-averaged benchmark score across all fixtures. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use in the main Results section when introducing the overall benchmark comparison.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.7523507864684335, "repetitions": 5, "stdDev": 0.0028950633451438217 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.7537982017982017, "repetitions": 5, "stdDev": 0.0024674185437360318 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.6876156355721573, "repetitions": 5, "stdDev": 0.022273540769297218 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.6488264834580625, "repetitions": 5, "stdDev": 0.03471337765983303 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.6741726708074535, "repetitions": 5, "stdDev": 0.009170103518359875 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "headline-duration", "title": "Headline session duration", "chartType": "bar", "scope": "config-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-duration", "caption": "Macro-averaged wall-clock session duration across benchmark fixtures. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing benchmark speed and operational latency.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 14758.100000000002, "repetitions": 5, "stdDev": 2758.3562605290863 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 53827.08, "repetitions": 5, "stdDev": 3035.6715512387036 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 27324.159999999996, "repetitions": 5, "stdDev": 753.3649202079949 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 37350.719999999994, "repetitions": 5, "stdDev": 2321.4753632119373 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 94820.59999999999, "repetitions": 5, "stdDev": 12273.105654030685 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 59318.09999999999, "repetitions": 5, "stdDev": 5135.255597825682 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "headline-total-tokens", "title": "Headline total tokens", "chartType": "bar", "scope": "config-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-total-tokens", "caption": "Macro-averaged total tokens per config. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing model context usage and inference footprint.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 66929.37999999999, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 51573.72, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 95968.95999999999, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 74240.1, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 56991.96, "repetitions": 5 } ] }, "talkingPoints": [ "Lower token usage generally means lower inference load for model runs." ] }, { "id": "headline-cost", "title": "Headline estimated cost", "chartType": "bar", "scope": "config-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-cost", "caption": "Macro-averaged estimated session cost in USD. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when comparing model-session costs.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.12494352, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.06276034999999999, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.35590119000000003, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.13221054699999996, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.08598008799999998, "repetitions": 5 } ] }, "talkingPoints": [ "Command runs are not model sessions, so their USD cost is shown as N/A." ] }, { "id": "headline-recall-precision", "title": "Headline recall and precision", "chartType": "grouped-bar", "scope": "config-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-recall-precision", "caption": "Macro-averaged recall and precision for find-vulns tasks.", "recommendedUse": "Use when explaining detection behavior: coverage of known vulnerabilities versus false-positive control.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.6822294372294373, "precision": 0.8979999999999999 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.6804112554112554, "precision": 0.9146666666666666 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.7138311688311687, "precision": 0.6956277056277057 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.8125974025974025, "precision": 0.5857619047619047 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.8089177489177489, "precision": 0.626021978021978 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "headline-score-vs-cost", "title": "Score vs estimated cost", "chartType": "scatter", "scope": "config-aggregate", "metric": "score-vs-cost", "unit": "number", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-score-vs-cost", "caption": "Model-only cost/quality tradeoff. Better points move toward the top-left: higher score at lower estimated session cost.", "recommendedUse": "Use as a Pareto-style tradeoff visual for model configs only.", "dataSummary": { "points": [ { "label": "Claude Opus 4.6 High", "runConfigType": "model", "x": 0.12494352, "y": 0.7523507864684335 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "x": 0.06276034999999999, "y": 0.7537982017982017 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "x": 0.35590119000000003, "y": 0.6876156355721573 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "x": 0.13221054699999996, "y": 0.6488264834580625 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "x": 0.08598008799999998, "y": 0.6741726708074535 } ] }, "talkingPoints": [ "Snyk Code SAST is excluded because command rows do not have comparable model-session cost.", "Claude Opus 4.6 Medium is the apparent dominant point in this tradeoff view." ], "xAxisLabel": "COST", "yAxisLabel": "SCORE", "xUnit": "usd", "yUnit": "percent" }, { "id": "headline-score-vs-duration", "title": "Score vs session duration", "chartType": "scatter", "scope": "config-aggregate", "metric": "score-vs-duration", "unit": "number", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-score-vs-duration", "caption": "Speed/quality tradeoff across model and command configs. Better points move toward the top-left: higher score in less wall-clock time.", "recommendedUse": "Use when comparing benchmark quality against elapsed runtime.", "dataSummary": { "points": [ { "label": "Snyk Code SAST", "runConfigType": "command", "x": 14758.100000000002, "y": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "x": 53827.08, "y": 0.7523507864684335 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "x": 27324.159999999996, "y": 0.7537982017982017 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "x": 37350.719999999994, "y": 0.6876156355721573 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "x": 94820.59999999999, "y": 0.6488264834580625 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "x": 59318.09999999999, "y": 0.6741726708074535 } ] }, "talkingPoints": [ "Snyk Code SAST is the apparent dominant point in this tradeoff view." ], "xAxisLabel": "DURATION", "yAxisLabel": "SCORE", "xUnit": "milliseconds", "yUnit": "percent" }, { "id": "headline-recall-vs-precision", "title": "Recall vs precision", "chartType": "scatter", "scope": "config-aggregate", "metric": "recall-vs-precision", "unit": "number", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-recall-vs-precision", "caption": "Detection tradeoff for find-vulns tasks. Better points move toward the top-right: more known vulnerabilities found with fewer false positives.", "recommendedUse": "Use when explaining whether configs are recall-oriented or precision-oriented.", "dataSummary": { "points": [ { "label": "Snyk Code SAST", "runConfigType": "command", "x": 1, "y": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "x": 0.6822294372294373, "y": 0.8979999999999999 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "x": 0.6804112554112554, "y": 0.9146666666666666 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "x": 0.7138311688311687, "y": 0.6956277056277057 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "x": 0.8125974025974025, "y": 0.5857619047619047 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "x": 0.8089177489177489, "y": 0.626021978021978 } ] }, "talkingPoints": [ "Snyk Code SAST is the apparent dominant point in this tradeoff view." ], "xAxisLabel": "RECALL", "yAxisLabel": "PRECISION", "xUnit": "percent", "yUnit": "percent" }, { "id": "headline-score-stability", "title": "Score stability", "chartType": "scatter", "scope": "config-aggregate", "metric": "score-stability", "unit": "number", "section": { "id": "headline", "title": "Headline comparison", "subtitle": "Macro-average across 10 fixtures", "kind": "headline" }, "placeholder": "", "htmlAnchor": "index.html#chart-headline-score-stability", "caption": "Quality and repeated-run stability. Better points move toward the top-left: higher score with lower score standard deviation.", "recommendedUse": "Use when discussing repeatability across benchmark repetitions.", "dataSummary": { "points": [ { "label": "Snyk Code SAST", "runConfigType": "command", "x": 0, "y": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "x": 0.0028950633451438217, "y": 0.7523507864684335 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "x": 0.0024674185437360318, "y": 0.7537982017982017 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "x": 0.022273540769297218, "y": 0.6876156355721573 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "x": 0.03471337765983303, "y": 0.6488264834580625 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "x": 0.009170103518359875, "y": 0.6741726708074535 } ] }, "talkingPoints": [ "A zero standard deviation for command-based SAST is meaningful for deterministic repeated runs.", "Snyk Code SAST is the apparent dominant point in this tradeoff view." ], "xAxisLabel": "SCORE SD", "yAxisLabel": "SCORE", "xUnit": "percent", "yUnit": "percent" }, { "id": "js-project-copperline-score", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-copperline-find-vulns", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities", "subtitle": "js-project-copperline-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-copperline-score", "caption": "Mean benchmark score for JS Snippet (Plugin Installer): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-copperline-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.45999999999999996, "repetitions": 5, "stdDev": 0.0547722557505166 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.5, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.4688888888888888, "repetitions": 5, "stdDev": 0.1262957532300695 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.6670695970695971, "repetitions": 5, "stdDev": 0.14978888209892435 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.7295238095238095, "repetitions": 5, "stdDev": 0.11963663807575012 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-copperline-duration", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-copperline-find-vulns", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities", "subtitle": "js-project-copperline-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-copperline-duration", "caption": "Mean wall-clock session duration for JS Snippet (Plugin Installer): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-copperline-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 12406.2, "repetitions": 5, "stdDev": 3204.0407612887825 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 31806, "repetitions": 5, "stdDev": 2965.26718863579 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 17509, "repetitions": 5, "stdDev": 611.176733850365 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 26269.2, "repetitions": 5, "stdDev": 6538.156674476377 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 62360.6, "repetitions": 5, "stdDev": 26221.988278923473 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 42939.2, "repetitions": 5, "stdDev": 4803.7614116440045 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-copperline-total-tokens", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-copperline-find-vulns", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities", "subtitle": "js-project-copperline-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-copperline-total-tokens", "caption": "Mean total tokens for JS Snippet (Plugin Installer): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-copperline-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 44432.6, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 41266, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 85404.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 59835.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 40984, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-copperline-cost", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-copperline-find-vulns", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities", "subtitle": "js-project-copperline-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-copperline-cost", "caption": "Mean estimated session cost in USD for JS Snippet (Plugin Installer): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-copperline-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.0623075, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.0477811, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.28766879999999995, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.08204037, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.06237602, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-copperline-recall-precision", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-copperline-find-vulns", "title": "JS Snippet (Plugin Installer): Find Vulnerabilities", "subtitle": "js-project-copperline-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-copperline-recall-precision", "caption": "Mean recall and precision for JS Snippet (Plugin Installer): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-copperline-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.3333333333333333, "precision": 0.8 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.3333333333333333, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.4666666666666666, "precision": 0.5666666666666667 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 1, "precision": 0.5157142857142857 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 0.5857142857142857 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-goldleaf-score", "title": "JS Snippet (Report Preview): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-goldleaf-find-vulns", "title": "JS Snippet (Report Preview): Find Vulnerabilities", "subtitle": "js-project-goldleaf-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-goldleaf-score", "caption": "Mean benchmark score for JS Snippet (Report Preview): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-goldleaf-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.6666666666666666, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.6666666666666666, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.5066666666666666, "repetitions": 5, "stdDev": 0.14605934866804426 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.39428571428571424, "repetitions": 5, "stdDev": 0.15930688876271556 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.330952380952381, "repetitions": 5, "stdDev": 0.09903159210993057 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-goldleaf-duration", "title": "JS Snippet (Report Preview): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-goldleaf-find-vulns", "title": "JS Snippet (Report Preview): Find Vulnerabilities", "subtitle": "js-project-goldleaf-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-goldleaf-duration", "caption": "Mean wall-clock session duration for JS Snippet (Report Preview): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-goldleaf-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 10858.4, "repetitions": 5, "stdDev": 1812.9301696425043 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 123046, "repetitions": 5, "stdDev": 30434.38628426734 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 26258.8, "repetitions": 5, "stdDev": 3145.35017128459 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 43341.8, "repetitions": 5, "stdDev": 5944.2820592566095 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 165321.4, "repetitions": 5, "stdDev": 33902.96924754527 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 85484.4, "repetitions": 5, "stdDev": 21836.984883449455 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-goldleaf-total-tokens", "title": "JS Snippet (Report Preview): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-goldleaf-find-vulns", "title": "JS Snippet (Report Preview): Find Vulnerabilities", "subtitle": "js-project-goldleaf-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-goldleaf-total-tokens", "caption": "Mean total tokens for JS Snippet (Report Preview): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-goldleaf-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 61091.6, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 41324.2, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 80647, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 76913.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 56886.4, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-goldleaf-cost", "title": "JS Snippet (Report Preview): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-goldleaf-find-vulns", "title": "JS Snippet (Report Preview): Find Vulnerabilities", "subtitle": "js-project-goldleaf-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-goldleaf-cost", "caption": "Mean estimated session cost in USD for JS Snippet (Report Preview): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-goldleaf-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.20126665, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.0441276, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.32484749999999996, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.19085059, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.10255824999999999, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-goldleaf-recall-precision", "title": "JS Snippet (Report Preview): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-goldleaf-find-vulns", "title": "JS Snippet (Report Preview): Find Vulnerabilities", "subtitle": "js-project-goldleaf-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-goldleaf-recall-precision", "caption": "Mean recall and precision for JS Snippet (Report Preview): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-goldleaf-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.5, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.5, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.5, "precision": 0.6 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.6, "precision": 0.29666666666666663 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.5, "precision": 0.2633333333333333 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-ironclad-score", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-ironclad-find-vulns", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities", "subtitle": "js-project-ironclad-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-ironclad-score", "caption": "Mean benchmark score for JS App (Knex/Postgres 3): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-ironclad-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.8528571428571429, "repetitions": 5, "stdDev": 0.09362321358749114 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.7976190476190476, "repetitions": 5, "stdDev": 0.08666797487238713 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.9047619047619048, "repetitions": 5, "stdDev": 0.14677176197545183 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-ironclad-duration", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-ironclad-find-vulns", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities", "subtitle": "js-project-ironclad-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-ironclad-duration", "caption": "Mean wall-clock session duration for JS App (Knex/Postgres 3): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-ironclad-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 14919.8, "repetitions": 5, "stdDev": 9610.178520714378 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 34464.8, "repetitions": 5, "stdDev": 1765.84319802184 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 26229, "repetitions": 5, "stdDev": 4714.936425870448 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 25555.6, "repetitions": 5, "stdDev": 3317.4838808952786 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 47295, "repetitions": 5, "stdDev": 5929.220817274391 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 38647.6, "repetitions": 5, "stdDev": 9070.4427841203 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-ironclad-total-tokens", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-ironclad-find-vulns", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities", "subtitle": "js-project-ironclad-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-ironclad-total-tokens", "caption": "Mean total tokens for JS App (Knex/Postgres 3): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-ironclad-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 53489.2, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 52213.6, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 83181.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 61618.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 44941.4, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-ironclad-cost", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-ironclad-find-vulns", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities", "subtitle": "js-project-ironclad-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-ironclad-cost", "caption": "Mean estimated session cost in USD for JS App (Knex/Postgres 3): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-ironclad-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.07569805, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.05450724999999999, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.24849435, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.06282675, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.04982358, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-ironclad-recall-precision", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-ironclad-find-vulns", "title": "JS App (Knex/Postgres 3): Find Vulnerabilities", "subtitle": "js-project-ironclad-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-ironclad-recall-precision", "caption": "Mean recall and precision for JS App (Knex/Postgres 3): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-ironclad-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.9333333333333332, "precision": 0.82 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 1, "precision": 0.67 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 0.85 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-nightowl-score", "title": "JS Todo App (SQLite 4): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-nightowl-find-vulns", "title": "JS Todo App (SQLite 4): Find Vulnerabilities", "subtitle": "js-project-nightowl-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-nightowl-score", "caption": "Mean benchmark score for JS Todo App (SQLite 4): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-nightowl-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.4, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.38545454545454544, "repetitions": 5, "stdDev": 0.019917183909278775 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.3065079365079365, "repetitions": 5, "stdDev": 0.07853614728156195 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.2942657342657342, "repetitions": 5, "stdDev": 0.04265466765276136 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.33904761904761904, "repetitions": 5, "stdDev": 0.08209280738860933 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-nightowl-duration", "title": "JS Todo App (SQLite 4): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-nightowl-find-vulns", "title": "JS Todo App (SQLite 4): Find Vulnerabilities", "subtitle": "js-project-nightowl-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-nightowl-duration", "caption": "Mean wall-clock session duration for JS Todo App (SQLite 4): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-nightowl-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 14569.2, "repetitions": 5, "stdDev": 4330.545369811983 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 72734.4, "repetitions": 5, "stdDev": 11991.31841375251 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 32466, "repetitions": 5, "stdDev": 989.8926709497348 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 62034.2, "repetitions": 5, "stdDev": 8485.879989724106 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 146828.6, "repetitions": 5, "stdDev": 48441.69560203276 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 75337.4, "repetitions": 5, "stdDev": 10468.581460732872 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-nightowl-total-tokens", "title": "JS Todo App (SQLite 4): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-nightowl-find-vulns", "title": "JS Todo App (SQLite 4): Find Vulnerabilities", "subtitle": "js-project-nightowl-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-nightowl-total-tokens", "caption": "Mean total tokens for JS Todo App (SQLite 4): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-nightowl-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 118891.4, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 77422.6, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 150744.8, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 120793.2, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 88261.6, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-nightowl-cost", "title": "JS Todo App (SQLite 4): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-nightowl-find-vulns", "title": "JS Todo App (SQLite 4): Find Vulnerabilities", "subtitle": "js-project-nightowl-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-nightowl-cost", "caption": "Mean estimated session cost in USD for JS Todo App (SQLite 4): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-nightowl-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.23883314999999997, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.0964772, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.6660503999999999, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.24272473, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.13626122999999998, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-nightowl-recall-precision", "title": "JS Todo App (SQLite 4): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-nightowl-find-vulns", "title": "JS Todo App (SQLite 4): Find Vulnerabilities", "subtitle": "js-project-nightowl-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-nightowl-recall-precision", "caption": "Mean recall and precision for JS Todo App (SQLite 4): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-nightowl-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.2857142857142857, "precision": 0.6666666666666666 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.2857142857142857, "precision": 0.5999999999999999 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.34285714285714286, "precision": 0.28145743145743146 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.2857142857142857, "precision": 0.31666666666666665 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.34285714285714286, "precision": 0.33571428571428574 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-purplehaze-score", "title": "JS Todo App (SQLite 5): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-purplehaze-find-vulns", "title": "JS Todo App (SQLite 5): Find Vulnerabilities", "subtitle": "js-project-purplehaze-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-purplehaze-score", "caption": "Mean benchmark score for JS Todo App (SQLite 5): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-purplehaze-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.5243137254901962, "repetitions": 5, "stdDev": 0.03853826670685816 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.5133333333333334, "repetitions": 5, "stdDev": 0.018257418583505474 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.7641294936947111, "repetitions": 5, "stdDev": 0.0644158441051645 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.691093117408907, "repetitions": 5, "stdDev": 0.06896627744159121 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.6936314699792961, "repetitions": 5, "stdDev": 0.09689889745585084 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-purplehaze-duration", "title": "JS Todo App (SQLite 5): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-purplehaze-find-vulns", "title": "JS Todo App (SQLite 5): Find Vulnerabilities", "subtitle": "js-project-purplehaze-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-purplehaze-duration", "caption": "Mean wall-clock session duration for JS Todo App (SQLite 5): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-purplehaze-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 13712, "repetitions": 5, "stdDev": 5506.103749476575 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 102118.8, "repetitions": 5, "stdDev": 9631.714655241818 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 49745.4, "repetitions": 5, "stdDev": 7422.845599903045 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 76464.4, "repetitions": 5, "stdDev": 9068.409248594817 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 207922.8, "repetitions": 5, "stdDev": 38458.793288401546 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 149816.6, "repetitions": 5, "stdDev": 28609.62684657037 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-purplehaze-total-tokens", "title": "JS Todo App (SQLite 5): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-purplehaze-find-vulns", "title": "JS Todo App (SQLite 5): Find Vulnerabilities", "subtitle": "js-project-purplehaze-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-purplehaze-total-tokens", "caption": "Mean total tokens for JS Todo App (SQLite 5): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-purplehaze-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 134109.6, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 64436.8, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 183588, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 112003.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 115114.2, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-purplehaze-cost", "title": "JS Todo App (SQLite 5): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-purplehaze-find-vulns", "title": "JS Todo App (SQLite 5): Find Vulnerabilities", "subtitle": "js-project-purplehaze-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-purplehaze-cost", "caption": "Mean estimated session cost in USD for JS Todo App (SQLite 5): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-purplehaze-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.30795005, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.1361562, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.78140565, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.29743083, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.23645153, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-purplehaze-recall-precision", "title": "JS Todo App (SQLite 5): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-purplehaze-find-vulns", "title": "JS Todo App (SQLite 5): Find Vulnerabilities", "subtitle": "js-project-purplehaze-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-purplehaze-recall-precision", "caption": "Mean recall and precision for JS Todo App (SQLite 5): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-purplehaze-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.38181818181818183, "precision": 0.8466666666666667 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.36363636363636365, "precision": 0.8800000000000001 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.7454545454545455, "precision": 0.7876767676767676 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.6545454545454545, "precision": 0.7533333333333333 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.7272727272727273, "precision": 0.6673626373626373 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-riverbend-score", "title": "JS Snippet (Import Profile): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-riverbend-find-vulns", "title": "JS Snippet (Import Profile): Find Vulnerabilities", "subtitle": "js-project-riverbend-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-riverbend-score", "caption": "Mean benchmark score for JS Snippet (Import Profile): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-riverbend-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.8666666666666666, "repetitions": 5, "stdDev": 0.18257418583505539 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.38999999999999996, "repetitions": 5, "stdDev": 0.0894427190999916 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.39333333333333337, "repetitions": 5, "stdDev": 0.0683130051063973 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-riverbend-duration", "title": "JS Snippet (Import Profile): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-riverbend-find-vulns", "title": "JS Snippet (Import Profile): Find Vulnerabilities", "subtitle": "js-project-riverbend-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-riverbend-duration", "caption": "Mean wall-clock session duration for JS Snippet (Import Profile): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-riverbend-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 26113.4, "repetitions": 5, "stdDev": 20389.500663331608 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 37061.4, "repetitions": 5, "stdDev": 1069.608479771921 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 21962.8, "repetitions": 5, "stdDev": 924.9547015935428 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 28835.4, "repetitions": 5, "stdDev": 3578.6462244821014 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 81866.6, "repetitions": 5, "stdDev": 17384.87977525298 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 41169.4, "repetitions": 5, "stdDev": 7776.343118715891 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-riverbend-total-tokens", "title": "JS Snippet (Import Profile): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-riverbend-find-vulns", "title": "JS Snippet (Import Profile): Find Vulnerabilities", "subtitle": "js-project-riverbend-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-riverbend-total-tokens", "caption": "Mean total tokens for JS Snippet (Import Profile): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-riverbend-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 50705, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 41214.8, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 79299.2, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 66077.2, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 44524, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-riverbend-cost", "title": "JS Snippet (Import Profile): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-riverbend-find-vulns", "title": "JS Snippet (Import Profile): Find Vulnerabilities", "subtitle": "js-project-riverbend-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-riverbend-cost", "caption": "Mean estimated session cost in USD for JS Snippet (Import Profile): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-riverbend-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.0705999, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.041769999999999995, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.24430230000000003, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.10130787, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.04947714, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-riverbend-recall-precision", "title": "JS Snippet (Import Profile): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-riverbend-find-vulns", "title": "JS Snippet (Import Profile): Find Vulnerabilities", "subtitle": "js-project-riverbend-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-riverbend-recall-precision", "caption": "Mean recall and precision for JS Snippet (Import Profile): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-riverbend-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 1, "precision": 0.8 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 1, "precision": 0.24523809523809526 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 0.24666666666666667 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-shadowfox-score", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-shadowfox-find-vulns", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities", "subtitle": "js-project-shadowfox-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-shadowfox-score", "caption": "Mean benchmark score for JS App (Knex/Postgres 2): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-shadowfox-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.8261904761904761, "repetitions": 5, "stdDev": 0.12587561797986957 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.8928571428571429, "repetitions": 5, "stdDev": 0.10714285714285719 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-shadowfox-duration", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-shadowfox-find-vulns", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities", "subtitle": "js-project-shadowfox-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-shadowfox-duration", "caption": "Mean wall-clock session duration for JS App (Knex/Postgres 2): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-shadowfox-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 10904.6, "repetitions": 5, "stdDev": 268.3361697572655 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 30168.2, "repetitions": 5, "stdDev": 1892.2211023027937 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 23024.6, "repetitions": 5, "stdDev": 2496.059855051557 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 23481, "repetitions": 5, "stdDev": 2516.4491848634657 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 57415, "repetitions": 5, "stdDev": 28661.505168082153 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 38718.2, "repetitions": 5, "stdDev": 13749.005662228816 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-shadowfox-total-tokens", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-shadowfox-find-vulns", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities", "subtitle": "js-project-shadowfox-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-shadowfox-total-tokens", "caption": "Mean total tokens for JS App (Knex/Postgres 2): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-shadowfox-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 52684, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 51544, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 76223.4, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 71383.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 52727.6, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-shadowfox-cost", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-shadowfox-find-vulns", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities", "subtitle": "js-project-shadowfox-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-shadowfox-cost", "caption": "Mean estimated session cost in USD for JS App (Knex/Postgres 2): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-shadowfox-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.06621269999999999, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.04721175, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.22242915000000002, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.11584166999999998, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.0686988, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-shadowfox-recall-precision", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-shadowfox-find-vulns", "title": "JS App (Knex/Postgres 2): Find Vulnerabilities", "subtitle": "js-project-shadowfox-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-shadowfox-recall-precision", "caption": "Mean recall and precision for JS App (Knex/Postgres 2): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-shadowfox-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 1, "precision": 0.72 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 0.82 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-silvergate-score", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-silvergate-find-vulns", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities", "subtitle": "js-project-silvergate-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-silvergate-score", "caption": "Mean benchmark score for JS Snippet (Redirect Handoff): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-silvergate-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.8571428571428571, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.8571428571428571, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.75, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.8425396825396826, "repetitions": 5, "stdDev": 0.12236849626767013 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.9428571428571428, "repetitions": 5, "stdDev": 0.07824607964359519 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-silvergate-duration", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-silvergate-find-vulns", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities", "subtitle": "js-project-silvergate-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-silvergate-duration", "caption": "Mean wall-clock session duration for JS Snippet (Redirect Handoff): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-silvergate-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 11899.8, "repetitions": 5, "stdDev": 3030.9397387609015 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 33903.6, "repetitions": 5, "stdDev": 1977.2450025224493 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 24686.8, "repetitions": 5, "stdDev": 1679.3018489836782 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 26038.4, "repetitions": 5, "stdDev": 2920.0436298110344 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 52957.6, "repetitions": 5, "stdDev": 13565.796744017654 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 39399.8, "repetitions": 5, "stdDev": 7233.165019547114 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-silvergate-total-tokens", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-silvergate-find-vulns", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities", "subtitle": "js-project-silvergate-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-silvergate-total-tokens", "caption": "Mean total tokens for JS Snippet (Redirect Handoff): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-silvergate-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 53035.6, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 41561.2, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 70881.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 56198.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 34455.2, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-silvergate-cost", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-silvergate-find-vulns", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities", "subtitle": "js-project-silvergate-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-silvergate-cost", "caption": "Mean estimated session cost in USD for JS Snippet (Redirect Handoff): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-silvergate-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.0734823, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.049429999999999995, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.23624415, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.06679653, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.04815783, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-silvergate-recall-precision", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-silvergate-find-vulns", "title": "JS Snippet (Redirect Handoff): Find Vulnerabilities", "subtitle": "js-project-silvergate-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-silvergate-recall-precision", "caption": "Mean recall and precision for JS Snippet (Redirect Handoff): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-silvergate-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.75, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.75, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.75, "precision": 0.75 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.9, "precision": 0.8133333333333332 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.9, "precision": 1 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-skylark-score", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-skylark-find-vulns", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities", "subtitle": "js-project-skylark-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-skylark-score", "caption": "Mean benchmark score for JS Snippet (Shelf Validator): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-skylark-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.7238095238095237, "repetitions": 5, "stdDev": 0.12777531299998796 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.8857142857142858, "repetitions": 5, "stdDev": 0.06388765649999402 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.8099999999999999, "repetitions": 5, "stdDev": 0.10839741694339404 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-skylark-duration", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-skylark-find-vulns", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities", "subtitle": "js-project-skylark-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-skylark-duration", "caption": "Mean wall-clock session duration for JS Snippet (Shelf Validator): Find Vulnerabilities. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-skylark-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 19728.4, "repetitions": 5, "stdDev": 16229.64233432148 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 38100.8, "repetitions": 5, "stdDev": 3330.3434057165937 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 28589.8, "repetitions": 5, "stdDev": 4130.742390902633 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 36131.6, "repetitions": 5, "stdDev": 5431.338352560996 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 73473.6, "repetitions": 5, "stdDev": 12533.49992220848 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 46073.2, "repetitions": 5, "stdDev": 2572.8253341414375 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-skylark-total-tokens", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-skylark-find-vulns", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities", "subtitle": "js-project-skylark-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-skylark-total-tokens", "caption": "Mean total tokens for JS Snippet (Shelf Validator): Find Vulnerabilities. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-skylark-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 56569.2, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 53388.4, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 87622.6, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 60812.8, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 43989.6, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-skylark-cost", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-skylark-find-vulns", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities", "subtitle": "js-project-skylark-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-skylark-cost", "caption": "Mean estimated session cost in USD for JS Snippet (Shelf Validator): Find Vulnerabilities. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-skylark-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.08526415000000001, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.0595753, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.32595540000000006, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.09176675, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.059187679999999986, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-skylark-recall-precision", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-skylark-find-vulns", "title": "JS Snippet (Shelf Validator): Find Vulnerabilities", "subtitle": "js-project-skylark-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-skylark-recall-precision", "caption": "Mean recall and precision for JS Snippet (Shelf Validator): Find Vulnerabilities.", "recommendedUse": "Use when discussing detection behavior for js-project-skylark-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.7999999999999999, "precision": 0.6666666666666666 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 1, "precision": 0.8 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.9333333333333333, "precision": 0.76 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] }, { "id": "js-project-tigerteam-score", "title": "JS App: Find Vulnerabilities 1 score", "chartType": "bar", "scope": "task-aggregate", "metric": "score", "unit": "percent", "section": { "id": "task-js-project-tigerteam-find-vulns", "title": "JS App: Find Vulnerabilities 1", "subtitle": "js-project-tigerteam-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-tigerteam-score", "caption": "Mean benchmark score for JS App: Find Vulnerabilities 1. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture performance for js-project-tigerteam-find-vulns.", "dataSummary": { "unit": "percent", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 1, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.6153846153846153, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.6153846153846153, "repetitions": 5, "stdDev": 0 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.6366300366300366, "repetitions": 5, "stdDev": 0.054969469545213416 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.6994871794871795, "repetitions": 5, "stdDev": 0.09411339279552379 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.7047619047619047, "repetitions": 5, "stdDev": 0.02129588549999802 } ] }, "talkingPoints": [ "Higher values are better.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-tigerteam-duration", "title": "JS App: Find Vulnerabilities 1 session duration", "chartType": "bar", "scope": "task-aggregate", "metric": "sessionDurationMs", "unit": "milliseconds", "section": { "id": "task-js-project-tigerteam-find-vulns", "title": "JS App: Find Vulnerabilities 1", "subtitle": "js-project-tigerteam-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-tigerteam-duration", "caption": "Mean wall-clock session duration for JS App: Find Vulnerabilities 1. Error bars show standard deviation across repeated runs.", "recommendedUse": "Use when discussing per-fixture runtime for js-project-tigerteam-find-vulns.", "dataSummary": { "unit": "milliseconds", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 12469.2, "repetitions": 5, "stdDev": 3048.827758335981 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 34866.8, "repetitions": 5, "stdDev": 5752.905587613966 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 22769.4, "repetitions": 5, "stdDev": 3470.3621280782786 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 25355.6, "repetitions": 5, "stdDev": 4110.66932506131 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 52764.8, "repetitions": 5, "stdDev": 10021.169178294516 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 35595.2, "repetitions": 5, "stdDev": 2225.622766777874 } ] }, "talkingPoints": [ "Lower duration is better for throughput.", "Repeated runs are summarized as mean plus standard deviation." ] }, { "id": "js-project-tigerteam-total-tokens", "title": "JS App: Find Vulnerabilities 1 total tokens", "chartType": "bar", "scope": "task-aggregate", "metric": "totalTokens", "unit": "tokens", "section": { "id": "task-js-project-tigerteam-find-vulns", "title": "JS App: Find Vulnerabilities 1", "subtitle": "js-project-tigerteam-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-tigerteam-total-tokens", "caption": "Mean total tokens for JS App: Find Vulnerabilities 1. Command-based SAST rows use zero-token accounting.", "recommendedUse": "Use when discussing per-fixture context usage for js-project-tigerteam-find-vulns.", "dataSummary": { "unit": "tokens", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": 0, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 44285.6, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 51365.6, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 62096.8, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 56764.2, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 48035.6, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-tigerteam-cost", "title": "JS App: Find Vulnerabilities 1 estimated cost", "chartType": "bar", "scope": "task-aggregate", "metric": "totalCostUsd", "unit": "usd", "section": { "id": "task-js-project-tigerteam-find-vulns", "title": "JS App: Find Vulnerabilities 1", "subtitle": "js-project-tigerteam-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-tigerteam-cost", "caption": "Mean estimated session cost in USD for JS App: Find Vulnerabilities 1. Command-based SAST rows report cost as not applicable.", "recommendedUse": "Use when discussing per-fixture model-session cost for js-project-tigerteam-find-vulns.", "dataSummary": { "unit": "usd", "rows": [ { "label": "Snyk Code SAST", "runConfigType": "command", "value": null, "repetitions": 5 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "value": 0.06782075000000001, "repetitions": 5 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "value": 0.0505671, "repetitions": 5 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "value": 0.2216142, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "value": 0.07051937999999999, "repetitions": 5 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "value": 0.04680882, "repetitions": 5 } ] }, "talkingPoints": [] }, { "id": "js-project-tigerteam-recall-precision", "title": "JS App: Find Vulnerabilities 1 recall and precision", "chartType": "grouped-bar", "scope": "task-aggregate", "metric": "recall-precision", "unit": "percent", "section": { "id": "task-js-project-tigerteam-find-vulns", "title": "JS App: Find Vulnerabilities 1", "subtitle": "js-project-tigerteam-find-vulns", "kind": "task" }, "placeholder": "", "htmlAnchor": "index.html#chart-js-project-tigerteam-recall-precision", "caption": "Mean recall and precision for JS App: Find Vulnerabilities 1.", "recommendedUse": "Use when discussing detection behavior for js-project-tigerteam-find-vulns.", "dataSummary": { "unit": "percent", "groups": [ { "label": "Snyk Code SAST", "runConfigType": "command", "recall": 1, "precision": 1 }, { "label": "Claude Opus 4.6 High", "runConfigType": "model", "recall": 0.5714285714285714, "precision": 0.6666666666666666 }, { "label": "Claude Opus 4.6 Medium", "runConfigType": "model", "recall": 0.5714285714285714, "precision": 0.6666666666666666 }, { "label": "Claude Opus 4.7 Max", "runConfigType": "model", "recall": 0.6, "precision": 0.6838095238095238 }, { "label": "Claude Sonnet 4.6 High", "runConfigType": "model", "recall": 0.6857142857142856, "precision": 0.7266666666666666 }, { "label": "Claude Sonnet 4.6 Medium", "runConfigType": "model", "recall": 0.6857142857142857, "precision": 0.7314285714285715 } ] }, "talkingPoints": [ "Recall measures known vulnerabilities found; precision measures how many reported findings matched ground truth.", "Higher recall and higher precision are both better." ] } ] }