opencode-swarm-plugin 0.39.1 → 0.42.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (82) hide show
  1. package/.hive/analysis/eval-failure-analysis-2025-12-25.md +331 -0
  2. package/.hive/analysis/session-data-quality-audit.md +320 -0
  3. package/.hive/eval-results.json +481 -24
  4. package/.hive/issues.jsonl +76 -11
  5. package/.hive/memories.jsonl +159 -1
  6. package/.opencode/eval-history.jsonl +315 -0
  7. package/.turbo/turbo-build.log +5 -5
  8. package/CHANGELOG.md +207 -0
  9. package/README.md +2 -0
  10. package/SCORER-ANALYSIS.md +598 -0
  11. package/bin/eval-gate.test.ts +158 -0
  12. package/bin/eval-gate.ts +74 -0
  13. package/bin/swarm.test.ts +1054 -719
  14. package/bin/swarm.ts +577 -0
  15. package/dist/compaction-hook.d.ts +10 -1
  16. package/dist/compaction-hook.d.ts.map +1 -1
  17. package/dist/compaction-observability.d.ts +173 -0
  18. package/dist/compaction-observability.d.ts.map +1 -0
  19. package/dist/compaction-prompt-scoring.d.ts +1 -0
  20. package/dist/compaction-prompt-scoring.d.ts.map +1 -1
  21. package/dist/eval-capture.d.ts +93 -0
  22. package/dist/eval-capture.d.ts.map +1 -1
  23. package/dist/eval-runner.d.ts +134 -0
  24. package/dist/eval-runner.d.ts.map +1 -0
  25. package/dist/hive.d.ts.map +1 -1
  26. package/dist/index.d.ts +65 -1
  27. package/dist/index.d.ts.map +1 -1
  28. package/dist/index.js +84043 -28070
  29. package/dist/memory-tools.d.ts +70 -2
  30. package/dist/memory-tools.d.ts.map +1 -1
  31. package/dist/memory.d.ts +37 -0
  32. package/dist/memory.d.ts.map +1 -1
  33. package/dist/observability-tools.d.ts +64 -0
  34. package/dist/observability-tools.d.ts.map +1 -1
  35. package/dist/plugin.js +83570 -27466
  36. package/dist/schemas/task.d.ts +3 -3
  37. package/dist/swarm-orchestrate.d.ts.map +1 -1
  38. package/dist/swarm-prompts.d.ts +32 -1
  39. package/dist/swarm-prompts.d.ts.map +1 -1
  40. package/docs/planning/ADR-009-oh-my-opencode-patterns.md +353 -0
  41. package/evals/ARCHITECTURE.md +1189 -0
  42. package/evals/README.md +113 -0
  43. package/evals/example.eval.ts +3 -4
  44. package/evals/fixtures/compaction-prompt-cases.ts +6 -0
  45. package/evals/scorers/coordinator-discipline.evalite-test.ts +163 -0
  46. package/evals/scorers/coordinator-discipline.ts +82 -2
  47. package/evals/scorers/index.test.ts +146 -0
  48. package/evals/scorers/index.ts +104 -0
  49. package/evals/swarm-decomposition.eval.ts +13 -4
  50. package/examples/commands/swarm.md +291 -21
  51. package/package.json +4 -3
  52. package/src/compaction-hook.ts +258 -110
  53. package/src/compaction-observability.integration.test.ts +139 -0
  54. package/src/compaction-observability.test.ts +187 -0
  55. package/src/compaction-observability.ts +324 -0
  56. package/src/compaction-prompt-scorers.test.ts +10 -9
  57. package/src/compaction-prompt-scoring.ts +7 -5
  58. package/src/eval-capture.test.ts +204 -1
  59. package/src/eval-capture.ts +194 -2
  60. package/src/eval-runner.test.ts +223 -0
  61. package/src/eval-runner.ts +402 -0
  62. package/src/hive.ts +57 -22
  63. package/src/index.ts +54 -1
  64. package/src/memory-tools.test.ts +84 -0
  65. package/src/memory-tools.ts +68 -3
  66. package/src/memory.test.ts +2 -2
  67. package/src/memory.ts +122 -49
  68. package/src/observability-tools.test.ts +13 -0
  69. package/src/observability-tools.ts +277 -0
  70. package/src/swarm-orchestrate.test.ts +162 -0
  71. package/src/swarm-orchestrate.ts +7 -5
  72. package/src/swarm-prompts.test.ts +168 -4
  73. package/src/swarm-prompts.ts +228 -7
  74. package/.env +0 -2
  75. package/.turbo/turbo-test.log +0 -481
  76. package/.turbo/turbo-typecheck.log +0 -1
  77. package/dist/beads.d.ts +0 -386
  78. package/dist/beads.d.ts.map +0 -1
  79. package/dist/schemas/bead-events.d.ts +0 -698
  80. package/dist/schemas/bead-events.d.ts.map +0 -1
  81. package/dist/schemas/bead.d.ts +0 -255
  82. package/dist/schemas/bead.d.ts.map +0 -1
@@ -0,0 +1,139 @@
1
+ /**
2
+ * Integration tests for compaction hook observability
3
+ *
4
+ * Tests the full integration of metrics collection with the compaction hook.
5
+ */
6
+
7
+ import { afterEach, beforeEach, describe, expect, it, mock } from "bun:test";
8
+ import { createCompactionHook } from "./compaction-hook";
9
+
10
+ // Track log calls
11
+ let logCalls: Array<{ level: string; data: unknown; message?: string }> = [];
12
+
13
+ const createMockLogger = () => ({
14
+ info: (data: unknown, message?: string) => {
15
+ logCalls.push({ level: "info", data, message });
16
+ },
17
+ debug: (data: unknown, message?: string) => {
18
+ logCalls.push({ level: "debug", data, message });
19
+ },
20
+ warn: (data: unknown, message?: string) => {
21
+ logCalls.push({ level: "warn", data, message });
22
+ },
23
+ error: (data: unknown, message?: string) => {
24
+ logCalls.push({ level: "error", data, message });
25
+ },
26
+ });
27
+
28
+ // Mock dependencies
29
+ mock.module("./hive", () => ({
30
+ getHiveWorkingDirectory: () => "/test/project",
31
+ getHiveAdapter: async () => ({
32
+ queryCells: async () => [],
33
+ }),
34
+ }));
35
+
36
+ mock.module("swarm-mail", () => ({
37
+ checkSwarmHealth: async () => ({
38
+ healthy: true,
39
+ database: "connected",
40
+ stats: {
41
+ events: 0,
42
+ agents: 0,
43
+ messages: 0,
44
+ reservations: 0,
45
+ },
46
+ }),
47
+ }));
48
+
49
+ mock.module("./logger", () => ({
50
+ createChildLogger: () => createMockLogger(),
51
+ }));
52
+
53
+ describe("Compaction Hook with Observability", () => {
54
+ beforeEach(() => {
55
+ logCalls = [];
56
+ });
57
+
58
+ afterEach(() => {
59
+ logCalls = [];
60
+ });
61
+
62
+ it("logs structured metrics on compaction run", async () => {
63
+ const hook = createCompactionHook();
64
+ const input = { sessionID: "test-session-123" };
65
+ const output = { context: [] as string[] };
66
+
67
+ await hook(input, output);
68
+
69
+ // Should have START log
70
+ const startLog = logCalls.find(
71
+ (log) => log.level === "info" && log.message === "compaction started",
72
+ );
73
+ expect(startLog).toBeDefined();
74
+ expect((startLog?.data as { session_id?: string })?.session_id).toBe("test-session-123");
75
+
76
+ // Should have COMPLETE log
77
+ const completeLog = logCalls.find(
78
+ (log) => log.level === "info" && log.message === "compaction complete",
79
+ );
80
+ expect(completeLog).toBeDefined();
81
+ expect((completeLog?.data as { duration_ms?: number })?.duration_ms).toBeGreaterThanOrEqual(0);
82
+ expect((completeLog?.data as { success?: boolean })?.success).toBe(true);
83
+ });
84
+
85
+ it("logs timing breakdown for each phase", async () => {
86
+ const hook = createCompactionHook();
87
+ await hook({ sessionID: "test" }, { context: [] });
88
+
89
+ // Should have debug logs for swarm-mail and hive checks
90
+ const swarmMailLog = logCalls.find(
91
+ (log) => log.level === "debug" && (log.data as { source?: string })?.source === "swarm-mail",
92
+ );
93
+ expect(swarmMailLog).toBeDefined();
94
+ expect((swarmMailLog?.data as { duration_ms?: number })?.duration_ms).toBeGreaterThanOrEqual(0);
95
+
96
+ const hiveLog = logCalls.find(
97
+ (log) => log.level === "debug" && (log.data as { source?: string })?.source === "hive",
98
+ );
99
+ expect(hiveLog).toBeDefined();
100
+ expect((hiveLog?.data as { duration_ms?: number })?.duration_ms).toBeGreaterThanOrEqual(0);
101
+ });
102
+
103
+ it("logs pattern extraction/skipping decisions", async () => {
104
+ const hook = createCompactionHook();
105
+ await hook({ sessionID: "test" }, { context: [] });
106
+
107
+ // Should have detection complete log
108
+ const detectionLog = logCalls.find(
109
+ (log) => log.level === "debug" && log.message === "swarm detection complete",
110
+ );
111
+ expect(detectionLog).toBeDefined();
112
+ expect((detectionLog?.data as { confidence?: string })?.confidence).toBeDefined();
113
+ });
114
+
115
+ it("captures metrics summary in completion log", async () => {
116
+ const hook = createCompactionHook();
117
+ await hook({ sessionID: "test-metrics" }, { context: [] });
118
+
119
+ const completeLog = logCalls.find(
120
+ (log) => log.level === "info" && log.message === "compaction complete",
121
+ );
122
+ expect(completeLog).toBeDefined();
123
+
124
+ const data = completeLog?.data as {
125
+ duration_ms?: number;
126
+ success?: boolean;
127
+ detected?: boolean;
128
+ confidence?: string;
129
+ context_injected?: boolean;
130
+ };
131
+
132
+ // Should have complete metrics
133
+ expect(data.duration_ms).toBeGreaterThanOrEqual(0);
134
+ expect(data.success).toBe(true);
135
+ expect(data.detected).toBeDefined();
136
+ expect(data.confidence).toBeDefined();
137
+ expect(data.context_injected).toBeDefined();
138
+ });
139
+ });
@@ -0,0 +1,187 @@
1
+ /**
2
+ * Tests for Compaction Observability
3
+ *
4
+ * TDD for adding structured metrics, timing breakdown, and queryable history
5
+ * to the pre-compaction hook.
6
+ */
7
+
8
+ import { describe, expect, it } from "bun:test";
9
+ import {
10
+ CompactionPhase,
11
+ createMetricsCollector,
12
+ getMetricsSummary,
13
+ recordPatternExtracted,
14
+ recordPatternSkipped,
15
+ recordPhaseComplete,
16
+ recordPhaseStart,
17
+ } from "./compaction-observability";
18
+
19
+ describe("CompactionMetrics", () => {
20
+ describe("Phase timing", () => {
21
+ it("tracks timing for each phase", () => {
22
+ const metrics = createMetricsCollector();
23
+
24
+ recordPhaseStart(metrics, CompactionPhase.DETECT);
25
+ // Simulate work
26
+ recordPhaseComplete(metrics, CompactionPhase.DETECT);
27
+
28
+ const summary = getMetricsSummary(metrics);
29
+
30
+ expect(summary.phases.DETECT).toBeDefined();
31
+ expect(summary.phases.DETECT.duration_ms).toBeGreaterThanOrEqual(0);
32
+ expect(summary.phases.DETECT.success).toBe(true);
33
+ });
34
+
35
+ it("tracks multiple phases independently", () => {
36
+ const metrics = createMetricsCollector();
37
+
38
+ recordPhaseStart(metrics, CompactionPhase.GATHER_SWARM_MAIL);
39
+ recordPhaseComplete(metrics, CompactionPhase.GATHER_SWARM_MAIL);
40
+
41
+ recordPhaseStart(metrics, CompactionPhase.GATHER_HIVE);
42
+ recordPhaseComplete(metrics, CompactionPhase.GATHER_HIVE);
43
+
44
+ const summary = getMetricsSummary(metrics);
45
+
46
+ expect(summary.phases.GATHER_SWARM_MAIL).toBeDefined();
47
+ expect(summary.phases.GATHER_HIVE).toBeDefined();
48
+ });
49
+
50
+ it("tracks phase failures", () => {
51
+ const metrics = createMetricsCollector();
52
+
53
+ recordPhaseStart(metrics, CompactionPhase.INJECT);
54
+ recordPhaseComplete(metrics, CompactionPhase.INJECT, {
55
+ success: false,
56
+ error: "Context injection failed"
57
+ });
58
+
59
+ const summary = getMetricsSummary(metrics);
60
+
61
+ expect(summary.phases.INJECT.success).toBe(false);
62
+ expect(summary.phases.INJECT.error).toBe("Context injection failed");
63
+ });
64
+ });
65
+
66
+ describe("Pattern extraction tracking", () => {
67
+ it("records extracted patterns with reasons", () => {
68
+ const metrics = createMetricsCollector();
69
+
70
+ recordPatternExtracted(metrics, "epic_state", "Epic bd-123 in_progress");
71
+ recordPatternExtracted(metrics, "agent_name", "BoldWind registered");
72
+
73
+ const summary = getMetricsSummary(metrics);
74
+
75
+ expect(summary.patterns_extracted).toBe(2);
76
+ expect(summary.extracted_patterns).toContain("epic_state");
77
+ expect(summary.extracted_patterns).toContain("agent_name");
78
+ });
79
+
80
+ it("records skipped patterns with reasons", () => {
81
+ const metrics = createMetricsCollector();
82
+
83
+ recordPatternSkipped(metrics, "subtask_details", "No subtasks found");
84
+
85
+ const summary = getMetricsSummary(metrics);
86
+
87
+ expect(summary.patterns_skipped).toBe(1);
88
+ expect(summary.skipped_patterns).toContain("subtask_details");
89
+ });
90
+
91
+ it("tracks extraction success rate", () => {
92
+ const metrics = createMetricsCollector();
93
+
94
+ recordPatternExtracted(metrics, "epic_state", "Found epic");
95
+ recordPatternExtracted(metrics, "agent_name", "Found agent");
96
+ recordPatternSkipped(metrics, "reservations", "None active");
97
+
98
+ const summary = getMetricsSummary(metrics);
99
+
100
+ expect(summary.patterns_extracted).toBe(2);
101
+ expect(summary.patterns_skipped).toBe(1);
102
+ expect(summary.extraction_success_rate).toBeCloseTo(0.67, 1);
103
+ });
104
+ });
105
+
106
+ describe("Overall metrics", () => {
107
+ it("tracks total duration from start to complete", () => {
108
+ const metrics = createMetricsCollector();
109
+ const startTime = Date.now();
110
+
111
+ recordPhaseStart(metrics, CompactionPhase.START);
112
+ recordPhaseComplete(metrics, CompactionPhase.START);
113
+
114
+ recordPhaseStart(metrics, CompactionPhase.DETECT);
115
+ recordPhaseComplete(metrics, CompactionPhase.DETECT);
116
+
117
+ recordPhaseStart(metrics, CompactionPhase.COMPLETE);
118
+ recordPhaseComplete(metrics, CompactionPhase.COMPLETE);
119
+
120
+ const summary = getMetricsSummary(metrics);
121
+
122
+ expect(summary.total_duration_ms).toBeGreaterThanOrEqual(0);
123
+ expect(summary.total_duration_ms).toBeLessThan(Date.now() - startTime + 10);
124
+ });
125
+
126
+ it("captures session metadata", () => {
127
+ const metrics = createMetricsCollector({
128
+ session_id: "test-session-123",
129
+ has_sdk_client: true,
130
+ });
131
+
132
+ const summary = getMetricsSummary(metrics);
133
+
134
+ expect(summary.session_id).toBe("test-session-123");
135
+ expect(summary.has_sdk_client).toBe(true);
136
+ });
137
+
138
+ it("captures final detection confidence", () => {
139
+ const metrics = createMetricsCollector();
140
+
141
+ recordPhaseComplete(metrics, CompactionPhase.DETECT, {
142
+ confidence: "high",
143
+ detected: true,
144
+ });
145
+
146
+ const summary = getMetricsSummary(metrics);
147
+
148
+ expect(summary.confidence).toBe("high");
149
+ expect(summary.detected).toBe(true);
150
+ });
151
+ });
152
+
153
+ describe("Debug mode", () => {
154
+ it("records verbose decision details when enabled", () => {
155
+ const metrics = createMetricsCollector({ debug: true });
156
+
157
+ recordPatternExtracted(metrics, "epic_state", "Epic bd-123 in_progress", {
158
+ cell_id: "bd-123",
159
+ cell_status: "in_progress",
160
+ subtask_count: 5,
161
+ });
162
+
163
+ const summary = getMetricsSummary(metrics);
164
+
165
+ // Debug details should be captured
166
+ expect(summary.debug_info).toBeDefined();
167
+ expect(summary.debug_info.length).toBeGreaterThan(0);
168
+ });
169
+ });
170
+ });
171
+
172
+ describe("Metrics persistence", () => {
173
+ it("can serialize metrics to JSON", () => {
174
+ const metrics = createMetricsCollector({ session_id: "test-123" });
175
+
176
+ recordPhaseStart(metrics, CompactionPhase.DETECT);
177
+ recordPhaseComplete(metrics, CompactionPhase.DETECT);
178
+ recordPatternExtracted(metrics, "epic_state", "Found epic");
179
+
180
+ const summary = getMetricsSummary(metrics);
181
+ const json = JSON.stringify(summary);
182
+
183
+ expect(json).toContain("test-123");
184
+ expect(json).toContain("DETECT");
185
+ expect(json).toContain("epic_state");
186
+ });
187
+ });
@@ -0,0 +1,324 @@
1
+ /**
2
+ * Compaction Hook Observability
3
+ *
4
+ * Structured logging, metrics, and queryable history for the pre-compaction hook.
5
+ *
6
+ * **Philosophy:** Make the invisible visible. When patterns aren't extracted,
7
+ * when detection fails, when timing explodes - we need to know WHY.
8
+ *
9
+ * @example
10
+ * ```typescript
11
+ * const metrics = createMetricsCollector({ session_id: "abc123" });
12
+ *
13
+ * recordPhaseStart(metrics, CompactionPhase.DETECT);
14
+ * // ... detection logic ...
15
+ * recordPhaseComplete(metrics, CompactionPhase.DETECT, { confidence: "high" });
16
+ *
17
+ * recordPatternExtracted(metrics, "epic_state", "Found epic bd-123");
18
+ *
19
+ * const summary = getMetricsSummary(metrics);
20
+ * console.log(`Detected: ${summary.detected}, Confidence: ${summary.confidence}`);
21
+ * ```
22
+ */
23
+
24
+ /**
25
+ * Compaction phases - aligned with existing log structure
26
+ *
27
+ * From compaction-hook.ts:
28
+ * - START: session_id, trigger
29
+ * - GATHER: source (swarm-mail|hive), duration_ms, stats/counts
30
+ * - DETECT: confidence, detected, reason_count, reasons
31
+ * - INJECT: confidence, context_length, context_type (full|fallback|none)
32
+ * - COMPLETE: duration_ms, success, detected, confidence, context_injected
33
+ */
34
+ export enum CompactionPhase {
35
+ START = "START",
36
+ GATHER_SWARM_MAIL = "GATHER_SWARM_MAIL",
37
+ GATHER_HIVE = "GATHER_HIVE",
38
+ DETECT = "DETECT",
39
+ INJECT = "INJECT",
40
+ COMPLETE = "COMPLETE",
41
+ }
42
+
43
+ /**
44
+ * Phase timing and outcome
45
+ */
46
+ interface PhaseMetrics {
47
+ duration_ms: number;
48
+ success: boolean;
49
+ error?: string;
50
+ /** Additional phase-specific data */
51
+ metadata?: Record<string, unknown>;
52
+ }
53
+
54
+ /**
55
+ * Pattern extraction record
56
+ */
57
+ interface PatternRecord {
58
+ pattern_type: string;
59
+ reason: string;
60
+ /** Debug details (only captured if debug mode enabled) */
61
+ details?: Record<string, unknown>;
62
+ timestamp: number;
63
+ }
64
+
65
+ /**
66
+ * Compaction metrics collector
67
+ *
68
+ * Mutable state object that accumulates metrics during a compaction run.
69
+ */
70
+ export interface CompactionMetrics {
71
+ /** Session metadata */
72
+ session_id?: string;
73
+ has_sdk_client?: boolean;
74
+ debug?: boolean;
75
+
76
+ /** Phase timings */
77
+ phases: Map<CompactionPhase, {
78
+ start_time: number;
79
+ end_time?: number;
80
+ metadata?: Record<string, unknown>;
81
+ error?: string;
82
+ }>;
83
+
84
+ /** Pattern extraction tracking */
85
+ extracted: PatternRecord[];
86
+ skipped: PatternRecord[];
87
+
88
+ /** Final detection result */
89
+ confidence?: "high" | "medium" | "low" | "none";
90
+ detected?: boolean;
91
+
92
+ /** Overall timing */
93
+ start_time: number;
94
+ end_time?: number;
95
+ }
96
+
97
+ /**
98
+ * Metrics summary (read-only snapshot)
99
+ */
100
+ export interface CompactionMetricsSummary {
101
+ session_id?: string;
102
+ has_sdk_client?: boolean;
103
+
104
+ /** Phase breakdown */
105
+ phases: Record<string, PhaseMetrics>;
106
+
107
+ /** Pattern extraction stats */
108
+ patterns_extracted: number;
109
+ patterns_skipped: number;
110
+ extraction_success_rate: number;
111
+ extracted_patterns: string[];
112
+ skipped_patterns: string[];
113
+
114
+ /** Detection outcome */
115
+ confidence?: "high" | "medium" | "low" | "none";
116
+ detected?: boolean;
117
+
118
+ /** Timing */
119
+ total_duration_ms: number;
120
+
121
+ /** Debug info (only if debug mode enabled) */
122
+ debug_info?: Array<{
123
+ phase: string;
124
+ pattern: string;
125
+ details: Record<string, unknown>;
126
+ }>;
127
+ }
128
+
129
+ /**
130
+ * Create a metrics collector
131
+ *
132
+ * @param metadata - Session metadata to capture
133
+ * @returns Mutable metrics collector
134
+ */
135
+ export function createMetricsCollector(metadata?: {
136
+ session_id?: string;
137
+ has_sdk_client?: boolean;
138
+ debug?: boolean;
139
+ }): CompactionMetrics {
140
+ return {
141
+ session_id: metadata?.session_id,
142
+ has_sdk_client: metadata?.has_sdk_client,
143
+ debug: metadata?.debug,
144
+ phases: new Map(),
145
+ extracted: [],
146
+ skipped: [],
147
+ start_time: Date.now(),
148
+ };
149
+ }
150
+
151
+ /**
152
+ * Record phase start
153
+ *
154
+ * @param metrics - Metrics collector
155
+ * @param phase - Phase being started
156
+ */
157
+ export function recordPhaseStart(
158
+ metrics: CompactionMetrics,
159
+ phase: CompactionPhase,
160
+ ): void {
161
+ metrics.phases.set(phase, {
162
+ start_time: Date.now(),
163
+ });
164
+ }
165
+
166
+ /**
167
+ * Record phase completion
168
+ *
169
+ * @param metrics - Metrics collector
170
+ * @param phase - Phase being completed
171
+ * @param result - Phase outcome
172
+ */
173
+ export function recordPhaseComplete(
174
+ metrics: CompactionMetrics,
175
+ phase: CompactionPhase,
176
+ result?: {
177
+ success?: boolean;
178
+ error?: string;
179
+ confidence?: "high" | "medium" | "low" | "none";
180
+ detected?: boolean;
181
+ [key: string]: unknown;
182
+ },
183
+ ): void {
184
+ let phaseData = metrics.phases.get(phase);
185
+ if (!phaseData) {
186
+ // Phase wasn't started, record it now
187
+ phaseData = {
188
+ start_time: Date.now(),
189
+ end_time: Date.now(),
190
+ error: "Phase completed without start",
191
+ };
192
+ metrics.phases.set(phase, phaseData);
193
+ } else {
194
+ phaseData.end_time = Date.now();
195
+ }
196
+
197
+ if (result) {
198
+ if (result.error) {
199
+ phaseData.error = result.error;
200
+ }
201
+
202
+ // Extract known fields
203
+ // biome-ignore lint/correctness/noUnusedVariables: extracting for spread
204
+ const { success, error, confidence, detected, ...rest } = result;
205
+
206
+ if (Object.keys(rest).length > 0) {
207
+ phaseData.metadata = rest;
208
+ }
209
+
210
+ // Update top-level detection result if provided in result
211
+ if (confidence) {
212
+ metrics.confidence = confidence;
213
+ }
214
+ if (detected !== undefined) {
215
+ metrics.detected = detected;
216
+ }
217
+ }
218
+ }
219
+
220
+ /**
221
+ * Record an extracted pattern
222
+ *
223
+ * @param metrics - Metrics collector
224
+ * @param pattern_type - Type of pattern extracted (e.g., "epic_state", "agent_name")
225
+ * @param reason - Human-readable reason for extraction
226
+ * @param details - Debug details (only captured if debug mode enabled)
227
+ */
228
+ export function recordPatternExtracted(
229
+ metrics: CompactionMetrics,
230
+ pattern_type: string,
231
+ reason: string,
232
+ details?: Record<string, unknown>,
233
+ ): void {
234
+ const record: PatternRecord = {
235
+ pattern_type,
236
+ reason,
237
+ timestamp: Date.now(),
238
+ };
239
+
240
+ if (metrics.debug && details) {
241
+ record.details = details;
242
+ }
243
+
244
+ metrics.extracted.push(record);
245
+ }
246
+
247
+ /**
248
+ * Record a skipped pattern
249
+ *
250
+ * @param metrics - Metrics collector
251
+ * @param pattern_type - Type of pattern that was skipped
252
+ * @param reason - Human-readable reason for skipping
253
+ */
254
+ export function recordPatternSkipped(
255
+ metrics: CompactionMetrics,
256
+ pattern_type: string,
257
+ reason: string,
258
+ ): void {
259
+ metrics.skipped.push({
260
+ pattern_type,
261
+ reason,
262
+ timestamp: Date.now(),
263
+ });
264
+ }
265
+
266
+ /**
267
+ * Get metrics summary (read-only snapshot)
268
+ *
269
+ * Computes derived metrics like success rates and total duration.
270
+ *
271
+ * @param metrics - Metrics collector
272
+ * @returns Immutable summary
273
+ */
274
+ export function getMetricsSummary(metrics: CompactionMetrics): CompactionMetricsSummary {
275
+ // Compute phase breakdown
276
+ const phases: Record<string, PhaseMetrics> = {};
277
+ for (const [phase, data] of metrics.phases) {
278
+ const duration = data.end_time ? data.end_time - data.start_time : 0;
279
+ phases[phase] = {
280
+ duration_ms: duration,
281
+ success: !data.error,
282
+ error: data.error,
283
+ metadata: data.metadata,
284
+ };
285
+ }
286
+
287
+ // Compute extraction stats
288
+ const totalPatterns = metrics.extracted.length + metrics.skipped.length;
289
+ const extractionSuccessRate = totalPatterns > 0
290
+ ? metrics.extracted.length / totalPatterns
291
+ : 0;
292
+
293
+ // Compute total duration
294
+ const totalDuration = metrics.end_time
295
+ ? metrics.end_time - metrics.start_time
296
+ : Date.now() - metrics.start_time;
297
+
298
+ const summary: CompactionMetricsSummary = {
299
+ session_id: metrics.session_id,
300
+ has_sdk_client: metrics.has_sdk_client,
301
+ phases,
302
+ patterns_extracted: metrics.extracted.length,
303
+ patterns_skipped: metrics.skipped.length,
304
+ extraction_success_rate: extractionSuccessRate,
305
+ extracted_patterns: metrics.extracted.map(p => p.pattern_type),
306
+ skipped_patterns: metrics.skipped.map(p => p.pattern_type),
307
+ confidence: metrics.confidence,
308
+ detected: metrics.detected,
309
+ total_duration_ms: totalDuration,
310
+ };
311
+
312
+ // Add debug info if enabled
313
+ if (metrics.debug) {
314
+ summary.debug_info = metrics.extracted
315
+ .filter(p => p.details !== undefined)
316
+ .map(p => ({
317
+ phase: "EXTRACT",
318
+ pattern: p.pattern_type,
319
+ details: p.details as Record<string, unknown>,
320
+ }));
321
+ }
322
+
323
+ return summary;
324
+ }
@@ -173,16 +173,17 @@ describe("forbiddenToolsPresent scorer", () => {
173
173
  - Edit (use swarm_spawn_subtask)
174
174
  - Write (use swarm_spawn_subtask)
175
175
  - swarmmail_reserve (only workers reserve)
176
- - bash with git commit (workers commit)`,
176
+ - git commit (workers commit)
177
+ - bash (for file modifications)`,
177
178
  };
178
179
 
179
180
  const result = scoreForbiddenToolsPresent(prompt);
180
181
 
181
182
  expect(result.score).toBe(1.0);
182
- expect(result.message).toContain("All 4 forbidden tools");
183
+ expect(result.message).toContain("All 5 forbidden tools");
183
184
  });
184
185
 
185
- test("scores 0.75 when 3 out of 4 tools listed", () => {
186
+ test("scores 0.6 when 3 out of 5 tools listed", () => {
186
187
  const prompt: CompactionPrompt = {
187
188
  content: `🚫 FORBIDDEN TOOLS:
188
189
  - Edit
@@ -192,19 +193,19 @@ describe("forbiddenToolsPresent scorer", () => {
192
193
 
193
194
  const result = scoreForbiddenToolsPresent(prompt);
194
195
 
195
- expect(result.score).toBe(0.75);
196
- expect(result.message).toContain("3/4");
196
+ expect(result.score).toBe(0.6);
197
+ expect(result.message).toContain("3/5");
197
198
  });
198
199
 
199
- test("scores 0.5 when 2 out of 4 tools listed", () => {
200
+ test("scores 0.4 when 2 out of 5 tools listed", () => {
200
201
  const prompt: CompactionPrompt = {
201
202
  content: `Don't use Edit or Write directly.`,
202
203
  };
203
204
 
204
205
  const result = scoreForbiddenToolsPresent(prompt);
205
206
 
206
- expect(result.score).toBe(0.5);
207
- expect(result.message).toContain("2/4");
207
+ expect(result.score).toBe(0.4);
208
+ expect(result.message).toContain("2/5");
208
209
  });
209
210
 
210
211
  test("scores 0.0 when no forbidden tools listed", () => {
@@ -215,7 +216,7 @@ describe("forbiddenToolsPresent scorer", () => {
215
216
  const result = scoreForbiddenToolsPresent(prompt);
216
217
 
217
218
  expect(result.score).toBe(0.0);
218
- expect(result.message).toContain("0/4");
219
+ expect(result.message).toContain("0/5");
219
220
  });
220
221
  });
221
222