auditor-lambda 0.2.8 → 0.2.10

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 (104) hide show
  1. package/README.md +6 -0
  2. package/audit-code-wrapper-lib.mjs +1 -1
  3. package/dist/adapters/eslint.js +9 -5
  4. package/dist/cli.d.ts +42 -1
  5. package/dist/cli.js +234 -63
  6. package/dist/extractors/bucketing.d.ts +4 -0
  7. package/dist/extractors/bucketing.js +6 -2
  8. package/dist/extractors/disposition.d.ts +4 -0
  9. package/dist/extractors/disposition.js +6 -2
  10. package/dist/extractors/fileInventory.js +24 -28
  11. package/dist/extractors/flows.d.ts +5 -0
  12. package/dist/extractors/flows.js +18 -38
  13. package/dist/extractors/pathPatterns.d.ts +10 -3
  14. package/dist/extractors/pathPatterns.js +109 -61
  15. package/dist/extractors/surfaces.d.ts +4 -0
  16. package/dist/extractors/surfaces.js +11 -11
  17. package/dist/index.d.ts +1 -1
  18. package/dist/index.js +2 -1
  19. package/dist/io/artifacts.d.ts +55 -40
  20. package/dist/io/artifacts.js +73 -110
  21. package/dist/io/json.js +52 -21
  22. package/dist/io/runArtifacts.d.ts +1 -1
  23. package/dist/io/runArtifacts.js +26 -3
  24. package/dist/orchestrator/advance.js +83 -62
  25. package/dist/orchestrator/autoFixExecutor.js +32 -15
  26. package/dist/orchestrator/flowCoverage.js +11 -5
  27. package/dist/orchestrator/flowPlanning.d.ts +7 -2
  28. package/dist/orchestrator/flowPlanning.js +46 -21
  29. package/dist/orchestrator/flowRequeue.js +28 -8
  30. package/dist/orchestrator/internalExecutors.js +12 -8
  31. package/dist/orchestrator/localCommands.d.ts +14 -0
  32. package/dist/orchestrator/localCommands.js +124 -0
  33. package/dist/orchestrator/planning.js +25 -3
  34. package/dist/orchestrator/requeue.js +11 -1
  35. package/dist/orchestrator/syntaxResolutionExecutor.js +60 -59
  36. package/dist/orchestrator/taskBuilder.d.ts +4 -2
  37. package/dist/orchestrator/taskBuilder.js +153 -52
  38. package/dist/orchestrator/unitBuilder.d.ts +3 -1
  39. package/dist/orchestrator/unitBuilder.js +24 -16
  40. package/dist/prompts/renderWorkerPrompt.d.ts +1 -1
  41. package/dist/prompts/renderWorkerPrompt.js +16 -8
  42. package/dist/providers/claudeCodeProvider.d.ts +4 -1
  43. package/dist/providers/claudeCodeProvider.js +8 -5
  44. package/dist/providers/localSubprocessProvider.d.ts +4 -0
  45. package/dist/providers/localSubprocessProvider.js +7 -2
  46. package/dist/providers/spawnLoggedCommand.d.ts +9 -1
  47. package/dist/providers/spawnLoggedCommand.js +77 -29
  48. package/dist/reporting/synthesis.d.ts +2 -0
  49. package/dist/reporting/synthesis.js +12 -9
  50. package/dist/supervisor/operatorHandoff.d.ts +1 -1
  51. package/dist/supervisor/operatorHandoff.js +56 -18
  52. package/dist/supervisor/runLedger.d.ts +1 -1
  53. package/dist/supervisor/runLedger.js +112 -5
  54. package/dist/supervisor/sessionConfig.js +10 -10
  55. package/dist/types/externalAnalyzer.d.ts +3 -0
  56. package/dist/types/flowCoverage.d.ts +5 -1
  57. package/dist/types/flowCoverage.js +5 -1
  58. package/dist/types/flows.d.ts +5 -1
  59. package/dist/types/flows.js +1 -1
  60. package/dist/types/runLedger.d.ts +5 -1
  61. package/dist/types/runLedger.js +6 -1
  62. package/dist/types/runtimeValidation.d.ts +12 -3
  63. package/dist/types/runtimeValidation.js +16 -1
  64. package/dist/types/sessionConfig.d.ts +15 -2
  65. package/dist/types/sessionConfig.js +15 -1
  66. package/dist/types/surfaces.d.ts +4 -1
  67. package/dist/types/surfaces.js +1 -1
  68. package/dist/types/workerSession.d.ts +9 -0
  69. package/dist/types/workerSession.js +5 -1
  70. package/dist/validation/artifacts.d.ts +1 -1
  71. package/dist/validation/artifacts.js +33 -20
  72. package/dist/validation/auditResults.d.ts +2 -2
  73. package/dist/validation/auditResults.js +7 -15
  74. package/dist/validation/basic.d.ts +9 -1
  75. package/dist/validation/basic.js +40 -3
  76. package/dist/validation/sessionConfig.d.ts +4 -2
  77. package/dist/validation/sessionConfig.js +62 -15
  78. package/docs/agent-integrations.md +31 -11
  79. package/docs/next-steps.md +21 -4
  80. package/docs/packaging.md +14 -0
  81. package/docs/product-direction.md +22 -0
  82. package/docs/production-launch-bar.md +2 -0
  83. package/docs/releasing.md +17 -0
  84. package/docs/remediation-baseline.md +75 -0
  85. package/docs/run-flow.md +23 -11
  86. package/docs/session-config.md +53 -6
  87. package/docs/supervisor.md +7 -0
  88. package/docs/workflow-refactor-brief.md +177 -0
  89. package/package.json +1 -1
  90. package/schemas/audit-code-v1alpha1.schema.json +1 -0
  91. package/schemas/audit_result.schema.json +4 -1
  92. package/schemas/audit_task.schema.json +3 -1
  93. package/schemas/coverage_matrix.schema.json +3 -3
  94. package/schemas/critical_flows.schema.json +6 -2
  95. package/schemas/file_disposition.schema.json +2 -2
  96. package/schemas/finding.schema.json +9 -4
  97. package/schemas/flow_coverage.schema.json +2 -2
  98. package/schemas/repo_manifest.schema.json +4 -4
  99. package/schemas/risk_register.schema.json +2 -2
  100. package/schemas/runtime_validation_report.schema.json +2 -2
  101. package/schemas/runtime_validation_tasks.schema.json +8 -2
  102. package/schemas/surface_manifest.schema.json +6 -3
  103. package/schemas/unit_manifest.schema.json +3 -2
  104. package/skills/audit-code/SKILL.md +5 -0
@@ -1,6 +1,9 @@
1
- import { requireKeys } from "./basic.js";
1
+ import { pushValidationIssue, requireKeys, } from "./basic.js";
2
2
  function pushIssue(issues, path, message) {
3
- issues.push({ path, message });
3
+ pushValidationIssue(issues, path, message);
4
+ }
5
+ function asArray(value) {
6
+ return Array.isArray(value) ? value : [];
4
7
  }
5
8
  export function validateArtifactBundle(bundle) {
6
9
  const issues = [];
@@ -37,14 +40,24 @@ export function validateArtifactBundle(bundle) {
37
40
  if (bundle.external_analyzer_results) {
38
41
  issues.push(...requireKeys(bundle.external_analyzer_results, "external_analyzer_results", ["tool", "results"]));
39
42
  }
40
- const repoPaths = new Set(bundle.repo_manifest?.files.map((file) => file.path) ?? []);
41
- const dispositionMap = new Map(bundle.file_disposition?.files.map((item) => [item.path, item.status]) ??
42
- []);
43
- const unitIds = new Set(bundle.unit_manifest?.units.map((unit) => unit.unit_id) ?? []);
44
- const flowIds = new Set(bundle.critical_flows?.flows.map((flow) => flow.id) ?? []);
45
- const runtimeTaskIds = new Set(bundle.runtime_validation_tasks?.tasks.map((task) => task.id) ?? []);
43
+ const repoManifestFiles = asArray(bundle.repo_manifest?.files);
44
+ const fileDispositionEntries = asArray(bundle.file_disposition?.files);
45
+ const unitManifestUnits = asArray(bundle.unit_manifest?.units);
46
+ const criticalFlows = asArray(bundle.critical_flows?.flows);
47
+ const flowCoverageEntries = asArray(bundle.flow_coverage?.flows);
48
+ const riskRegisterItems = asArray(bundle.risk_register?.items);
49
+ const surfaceEntries = asArray(bundle.surface_manifest?.surfaces);
50
+ const runtimeValidationTasks = asArray(bundle.runtime_validation_tasks?.tasks);
51
+ const runtimeValidationResults = asArray(bundle.runtime_validation_report?.results);
52
+ const externalAnalyzerResults = asArray(bundle.external_analyzer_results?.results);
53
+ const coverageFiles = asArray(bundle.coverage_matrix?.files);
54
+ const repoPaths = new Set(repoManifestFiles.map((file) => file.path));
55
+ const dispositionMap = new Map(fileDispositionEntries.map((item) => [item.path, item.status]));
56
+ const unitIds = new Set(unitManifestUnits.map((unit) => unit.unit_id));
57
+ const flowIds = new Set(criticalFlows.map((flow) => flow.id));
58
+ const runtimeTaskIds = new Set(runtimeValidationTasks.map((task) => task.id));
46
59
  if (bundle.repo_manifest && bundle.coverage_matrix) {
47
- const coveragePaths = new Set(bundle.coverage_matrix.files.map((file) => file.path));
60
+ const coveragePaths = new Set(coverageFiles.map((file) => file.path));
48
61
  for (const path of repoPaths) {
49
62
  if (!coveragePaths.has(path)) {
50
63
  pushIssue(issues, "coverage_matrix", `Missing coverage entry for ${path}`);
@@ -52,7 +65,7 @@ export function validateArtifactBundle(bundle) {
52
65
  }
53
66
  }
54
67
  if (bundle.repo_manifest && bundle.file_disposition) {
55
- const dispositionPaths = new Set(bundle.file_disposition.files.map((file) => file.path));
68
+ const dispositionPaths = new Set(fileDispositionEntries.map((file) => file.path));
56
69
  for (const path of repoPaths) {
57
70
  if (!dispositionPaths.has(path)) {
58
71
  pushIssue(issues, "file_disposition", `Missing disposition entry for ${path}`);
@@ -60,7 +73,7 @@ export function validateArtifactBundle(bundle) {
60
73
  }
61
74
  }
62
75
  if (bundle.unit_manifest) {
63
- for (const unit of bundle.unit_manifest.units) {
76
+ for (const unit of unitManifestUnits) {
64
77
  if (unit.files.length === 0) {
65
78
  pushIssue(issues, `unit_manifest:${unit.unit_id}`, "Unit has no files");
66
79
  }
@@ -79,7 +92,7 @@ export function validateArtifactBundle(bundle) {
79
92
  }
80
93
  }
81
94
  if (bundle.coverage_matrix && bundle.unit_manifest) {
82
- for (const file of bundle.coverage_matrix.files) {
95
+ for (const file of coverageFiles) {
83
96
  if (!repoPaths.has(file.path)) {
84
97
  pushIssue(issues, "coverage_matrix", `Coverage contains unknown file ${file.path}`);
85
98
  }
@@ -103,7 +116,7 @@ export function validateArtifactBundle(bundle) {
103
116
  }
104
117
  }
105
118
  if (bundle.critical_flows) {
106
- for (const flow of bundle.critical_flows.flows) {
119
+ for (const flow of criticalFlows) {
107
120
  if (flow.paths.length === 0) {
108
121
  pushIssue(issues, `critical_flows:${flow.id}`, "Flow has no paths");
109
122
  }
@@ -122,7 +135,7 @@ export function validateArtifactBundle(bundle) {
122
135
  }
123
136
  }
124
137
  if (bundle.flow_coverage && bundle.critical_flows) {
125
- for (const flow of bundle.flow_coverage.flows) {
138
+ for (const flow of flowCoverageEntries) {
126
139
  if (!flowIds.has(flow.flow_id)) {
127
140
  pushIssue(issues, `flow_coverage:${flow.flow_id}`, `Flow coverage references unknown flow ${flow.flow_id}`);
128
141
  }
@@ -143,15 +156,15 @@ export function validateArtifactBundle(bundle) {
143
156
  }
144
157
  }
145
158
  if (bundle.risk_register && bundle.unit_manifest) {
146
- const riskUnitIds = new Set(bundle.risk_register.items.map((item) => item.unit_id));
147
- for (const unit of bundle.unit_manifest.units) {
159
+ const riskUnitIds = new Set(riskRegisterItems.map((item) => item.unit_id));
160
+ for (const unit of unitManifestUnits) {
148
161
  if (!riskUnitIds.has(unit.unit_id)) {
149
162
  pushIssue(issues, "risk_register", `Missing risk entry for unit ${unit.unit_id}`);
150
163
  }
151
164
  }
152
165
  }
153
166
  if (bundle.surface_manifest) {
154
- for (const surface of bundle.surface_manifest.surfaces) {
167
+ for (const surface of surfaceEntries) {
155
168
  if (!repoPaths.has(surface.entrypoint)) {
156
169
  pushIssue(issues, `surface_manifest:${surface.id}`, `Surface references unknown entrypoint ${surface.entrypoint}`);
157
170
  }
@@ -162,7 +175,7 @@ export function validateArtifactBundle(bundle) {
162
175
  }
163
176
  }
164
177
  if (bundle.runtime_validation_tasks) {
165
- for (const task of bundle.runtime_validation_tasks.tasks) {
178
+ for (const task of runtimeValidationTasks) {
166
179
  if (task.target_paths.length === 0) {
167
180
  pushIssue(issues, `runtime_validation_tasks:${task.id}`, "Runtime validation task has no target paths");
168
181
  }
@@ -174,14 +187,14 @@ export function validateArtifactBundle(bundle) {
174
187
  }
175
188
  }
176
189
  if (bundle.runtime_validation_report) {
177
- for (const result of bundle.runtime_validation_report.results) {
190
+ for (const result of runtimeValidationResults) {
178
191
  if (!runtimeTaskIds.has(result.task_id)) {
179
192
  pushIssue(issues, `runtime_validation_report:${result.task_id}`, `Runtime validation result references unknown task ${result.task_id}`);
180
193
  }
181
194
  }
182
195
  }
183
196
  if (bundle.external_analyzer_results) {
184
- for (const item of bundle.external_analyzer_results.results) {
197
+ for (const item of externalAnalyzerResults) {
185
198
  if (!repoPaths.has(item.path) && bundle.repo_manifest) {
186
199
  pushIssue(issues, `external_analyzer_results:${item.id}`, `External analyzer result references unknown path ${item.path}`);
187
200
  }
@@ -1,11 +1,11 @@
1
1
  import type { AuditTask } from "../types.js";
2
+ import { type ValidationIssue } from "./basic.js";
2
3
  export type IssueSeverity = "error" | "warning";
3
- export interface AuditResultIssue {
4
+ export interface AuditResultIssue extends ValidationIssue {
4
5
  result_index: number;
5
6
  task_id: string;
6
7
  severity: IssueSeverity;
7
8
  field: string;
8
- message: string;
9
9
  }
10
10
  export interface ValidateAuditResultOptions {
11
11
  lineIndex?: Record<string, number>;
@@ -1,3 +1,4 @@
1
+ import { describeValue, formatValidationIssues, isRecord, } from "./basic.js";
1
2
  const REQUIRED_FINDING_FIELDS = [
2
3
  "id",
3
4
  "title",
@@ -24,21 +25,10 @@ const VALID_LENSES = new Set([
24
25
  function pushIssue(issues, params) {
25
26
  issues.push({
26
27
  ...params,
28
+ path: params.path ?? params.field,
27
29
  severity: params.severity ?? "error",
28
30
  });
29
31
  }
30
- function describeValue(value) {
31
- if (Array.isArray(value)) {
32
- return "array";
33
- }
34
- if (value === null) {
35
- return "null";
36
- }
37
- return typeof value;
38
- }
39
- function isRecord(value) {
40
- return typeof value === "object" && value !== null && !Array.isArray(value);
41
- }
42
32
  function isNonEmptyString(value) {
43
33
  return typeof value === "string" && value.trim().length > 0;
44
34
  }
@@ -404,7 +394,9 @@ export function validateAuditResults(results, tasks, options = {}) {
404
394
  return issues;
405
395
  }
406
396
  export function formatAuditResultIssues(issues) {
407
- return issues
408
- .map((issue) => ` [${issue.severity}] ${issue.task_id} / ${issue.field}: ${issue.message}`)
409
- .join("\n");
397
+ return formatValidationIssues(issues.map((issue) => ({
398
+ path: `${issue.task_id} / ${issue.field}`,
399
+ message: issue.message,
400
+ severity: issue.severity,
401
+ })));
410
402
  }
@@ -1,5 +1,13 @@
1
+ export type ValidationSeverity = "error" | "warning";
1
2
  export interface ValidationIssue {
2
3
  path: string;
3
4
  message: string;
5
+ severity: ValidationSeverity;
4
6
  }
5
- export declare function requireKeys(record: Record<string, unknown>, path: string, keys: string[]): ValidationIssue[];
7
+ export declare function describeValue(value: unknown): string;
8
+ export declare function isRecord(value: unknown): value is Record<string, unknown>;
9
+ export declare function createValidationIssue(path: string, message: string, severity?: ValidationSeverity): ValidationIssue;
10
+ export declare function pushValidationIssue(issues: ValidationIssue[], path: string, message: string, severity?: ValidationSeverity): void;
11
+ export declare function prefixValidationIssues(prefix: string, issues: ValidationIssue[]): ValidationIssue[];
12
+ export declare function formatValidationIssues(issues: ValidationIssue[]): string;
13
+ export declare function requireKeys(value: unknown, path: string, keys: readonly string[]): ValidationIssue[];
@@ -1,8 +1,45 @@
1
- export function requireKeys(record, path, keys) {
1
+ export function describeValue(value) {
2
+ if (Array.isArray(value)) {
3
+ return "array";
4
+ }
5
+ if (value === null) {
6
+ return "null";
7
+ }
8
+ return typeof value;
9
+ }
10
+ export function isRecord(value) {
11
+ return typeof value === "object" && value !== null && !Array.isArray(value);
12
+ }
13
+ export function createValidationIssue(path, message, severity = "error") {
14
+ return { path, message, severity };
15
+ }
16
+ export function pushValidationIssue(issues, path, message, severity = "error") {
17
+ issues.push(createValidationIssue(path, message, severity));
18
+ }
19
+ export function prefixValidationIssues(prefix, issues) {
20
+ return issues.map((issue) => ({
21
+ ...issue,
22
+ path: issue.path.length === 0
23
+ ? prefix
24
+ : issue.path === prefix || issue.path.startsWith(`${prefix}.`)
25
+ ? issue.path
26
+ : `${prefix}.${issue.path}`,
27
+ }));
28
+ }
29
+ export function formatValidationIssues(issues) {
30
+ return issues
31
+ .map((issue) => ` [${issue.severity}] ${issue.path}: ${issue.message}`)
32
+ .join("\n");
33
+ }
34
+ export function requireKeys(value, path, keys) {
2
35
  const issues = [];
36
+ if (!isRecord(value)) {
37
+ pushValidationIssue(issues, path, `Expected an object, got ${describeValue(value)}.`);
38
+ return issues;
39
+ }
3
40
  for (const key of keys) {
4
- if (!(key in record)) {
5
- issues.push({ path, message: `Missing required key: ${key}` });
41
+ if (!(key in value)) {
42
+ pushValidationIssue(issues, path, `Missing required key: ${key}`);
6
43
  }
7
44
  }
8
45
  return issues;
@@ -1,6 +1,8 @@
1
- import type { SessionConfig } from "../types/sessionConfig.js";
2
- import type { ValidationIssue } from "./basic.js";
1
+ import { type SessionConfig } from "../types/sessionConfig.js";
2
+ import { type ValidationIssue } from "./basic.js";
3
3
  export declare function validateSessionConfig(value: unknown): ValidationIssue[];
4
4
  export declare function validateConfiguredProviderEnvironment(sessionConfig: SessionConfig, options?: {
5
5
  commandExists?: (command: string) => boolean;
6
+ pathExists?: (commandPath: string) => boolean;
6
7
  }): ValidationIssue[];
8
+ export { formatValidationIssues } from "./basic.js";
@@ -1,18 +1,11 @@
1
1
  import { spawnSync } from "node:child_process";
2
- const VALID_PROVIDERS = new Set([
3
- "auto",
4
- "local-subprocess",
5
- "subprocess-template",
6
- "claude-code",
7
- "opencode",
8
- "vscode-task",
9
- ]);
10
- const VALID_UI_MODES = new Set(["headless", "visible"]);
2
+ import { accessSync, constants } from "node:fs";
3
+ import { PROVIDER_NAMES, SESSION_UI_MODES, } from "../types/sessionConfig.js";
4
+ import { isRecord, pushValidationIssue, } from "./basic.js";
5
+ const VALID_PROVIDERS = new Set(PROVIDER_NAMES);
6
+ const VALID_UI_MODES = new Set(SESSION_UI_MODES);
11
7
  function pushIssue(issues, path, message) {
12
- issues.push({ path, message });
13
- }
14
- function isRecord(value) {
15
- return typeof value === "object" && value !== null && !Array.isArray(value);
8
+ pushValidationIssue(issues, path, message);
16
9
  }
17
10
  function validateStringArray(value, path, label, issues, options = {}) {
18
11
  if (!Array.isArray(value)) {
@@ -74,6 +67,9 @@ function validateAgentProviderSection(value, path, issues) {
74
67
  if (typeof value.command !== "string" || value.command.trim().length === 0) {
75
68
  pushIssue(issues, `${path}.command`, "command must be a non-empty string when provided.");
76
69
  }
70
+ else if (!isSupportedConfiguredCommand(value.command)) {
71
+ pushIssue(issues, `${path}.command`, "command must be a bare executable name or direct executable path. Put CLI flags in extra_args.");
72
+ }
77
73
  }
78
74
  if (value.extra_args !== undefined) {
79
75
  validateStringArray(value.extra_args, `${path}.extra_args`, "extra_args", issues, { allowEmptyArray: true });
@@ -84,6 +80,43 @@ function commandExists(command) {
84
80
  const result = spawnSync(lookupCommand, [command], { stdio: "ignore" });
85
81
  return result.status === 0;
86
82
  }
83
+ function configuredPathExists(commandPath) {
84
+ try {
85
+ accessSync(commandPath, constants.F_OK);
86
+ return true;
87
+ }
88
+ catch {
89
+ return false;
90
+ }
91
+ }
92
+ function startsWithPathPrefix(command) {
93
+ return (command.startsWith(".") ||
94
+ command.startsWith("/") ||
95
+ command.startsWith("\\\\") ||
96
+ /^[A-Za-z]:[\\/]/.test(command));
97
+ }
98
+ function containsForbiddenCommandSyntax(command) {
99
+ return /[\r\n"'`|&;<>]/.test(command);
100
+ }
101
+ function isBareExecutableName(command) {
102
+ return (command.length > 0 &&
103
+ !/\s/.test(command) &&
104
+ !containsForbiddenCommandSyntax(command) &&
105
+ !/[\\/]/.test(command) &&
106
+ !/^[A-Za-z]:/.test(command));
107
+ }
108
+ function isDirectExecutablePath(command) {
109
+ return (command.length > 0 &&
110
+ !containsForbiddenCommandSyntax(command) &&
111
+ startsWithPathPrefix(command));
112
+ }
113
+ function isSupportedConfiguredCommand(command) {
114
+ const trimmed = command.trim();
115
+ if (trimmed.length === 0 || trimmed !== command) {
116
+ return false;
117
+ }
118
+ return isBareExecutableName(trimmed) || isDirectExecutablePath(trimmed);
119
+ }
87
120
  export function validateSessionConfig(value) {
88
121
  const issues = [];
89
122
  if (value === undefined) {
@@ -122,18 +155,32 @@ export function validateSessionConfig(value) {
122
155
  export function validateConfiguredProviderEnvironment(sessionConfig, options = {}) {
123
156
  const issues = [];
124
157
  const lookupCommand = options.commandExists ?? commandExists;
158
+ const lookupPath = options.pathExists ?? configuredPathExists;
125
159
  const provider = sessionConfig.provider ?? "local-subprocess";
126
160
  if (provider === "claude-code") {
127
161
  const command = sessionConfig.claude_code?.command ?? "claude";
128
- if (!lookupCommand(command)) {
162
+ if (isBareExecutableName(command) && !lookupCommand(command)) {
129
163
  pushIssue(issues, "claude_code.command", `Configured claude-code executable was not found on PATH: ${command}.`);
130
164
  }
165
+ else if (isDirectExecutablePath(command) && !lookupPath(command)) {
166
+ pushIssue(issues, "claude_code.command", `Configured claude-code executable path does not exist: ${command}.`);
167
+ }
168
+ else if (!isSupportedConfiguredCommand(command)) {
169
+ pushIssue(issues, "claude_code.command", "Configured claude-code command must be a bare executable name or direct path. Put CLI flags in extra_args.");
170
+ }
131
171
  }
132
172
  if (provider === "opencode") {
133
173
  const command = sessionConfig.opencode?.command ?? "opencode";
134
- if (!lookupCommand(command)) {
174
+ if (isBareExecutableName(command) && !lookupCommand(command)) {
135
175
  pushIssue(issues, "opencode.command", `Configured opencode executable was not found on PATH: ${command}.`);
136
176
  }
177
+ else if (isDirectExecutablePath(command) && !lookupPath(command)) {
178
+ pushIssue(issues, "opencode.command", `Configured opencode executable path does not exist: ${command}.`);
179
+ }
180
+ else if (!isSupportedConfiguredCommand(command)) {
181
+ pushIssue(issues, "opencode.command", "Configured opencode command must be a bare executable name or direct path. Put CLI flags in extra_args.");
182
+ }
137
183
  }
138
184
  return issues;
139
185
  }
186
+ export { formatValidationIssues } from "./basic.js";
@@ -14,6 +14,17 @@ Normal product usage should:
14
14
  - avoid manual `--root`, provider flags, and model selection in normal use
15
15
  - let the supervisor advance the audit automatically until it completes or no further automatic progress is possible
16
16
 
17
+ ## Review ownership rule
18
+
19
+ Semantic review should stay with the active conversation agent by default.
20
+
21
+ That means:
22
+
23
+ - use the current host conversation as the normal owner of review work
24
+ - if the host agent can delegate to subagents in parallel, let the host runtime make that decision
25
+ - do not treat `.audit-artifacts/session-config.json` as the normal way to choose a second LLM for review
26
+ - treat backend provider adapters as compatibility bridges for fallback CLI usage only
27
+
17
28
  ## Conversation-first setup
18
29
 
19
30
  The canonical prompt asset is:
@@ -136,7 +147,11 @@ Terminal interpretation:
136
147
  - `audit_state.status === "complete"` means the audit finished end to end.
137
148
  - `audit_state.status === "blocked"` means the wrapper exhausted automatic work and the remaining review still needs imported results or a provider-capable continuation path.
138
149
 
139
- When `provider` is configured as `claude-code`, `opencode`, `subprocess-template`, or `vscode-task`, the wrapper can now continue through audit-task review in the same invocation as long as that provider can write structured `AuditResult[]` output and hand control back to the bounded worker command.
150
+ Current implementation note:
151
+
152
+ - the backend fallback still supports explicit provider bridges such as `claude-code`, `opencode`, `subprocess-template`, and `vscode-task`
153
+ - those bridges are compatibility modes, not the intended default review owner
154
+ - the intended long-term workflow is documented in [docs/workflow-refactor-brief.md](/C:/Code/auditor-lambda/docs/workflow-refactor-brief.md)
140
155
 
141
156
  When additional evidence exists, pass it into the same wrapper:
142
157
 
@@ -149,6 +164,7 @@ audit-code --external-analyzer-results /path/to/external_analyzer_results.json
149
164
  Each response also refreshes `.audit-artifacts/operator-handoff.json` and `.audit-artifacts/operator-handoff.md` so operators can see the pending obligations, suggested import paths, and session-config continuation hint without reconstructing the state manually.
150
165
 
151
166
  Everything below is backend fallback guidance, not the primary product path.
167
+ Use it when the current host cannot keep review inside the active conversation, not as the first choice for semantic-review ownership.
152
168
 
153
169
  ## Provider matrix
154
170
 
@@ -156,9 +172,9 @@ Everything below is backend fallback guidance, not the primary product path.
156
172
 
157
173
  Use when you want the supervisor to stay entirely local.
158
174
 
159
- This requires no external agent CLI. The supervisor launches fresh worker subprocesses that call the bounded `worker-run` entrypoint for deterministic stages.
175
+ This requires no external agent CLI. Deterministic executors run in-process during normal wrapper runs, and the supervisor only stops once the remaining work is genuinely semantic review.
160
176
 
161
- When the remaining work is genuinely audit-task review, `local-subprocess` stops in a terminal blocked handoff instead of pretending more automatic progress is available.
177
+ When that review boundary is reached, `local-subprocess` stops in a terminal blocked handoff instead of pretending more automatic progress is available. Use `--results <file>` for a single batch or `--batch-results <dir>` when the active conversation agent reviewed multiple task batches before ingestion.
162
178
 
163
179
  This is the safest default backend when the repository is already available locally.
164
180
 
@@ -166,19 +182,17 @@ This is the safest default backend when the repository is already available loca
166
182
 
167
183
  Use when Claude Code is installed and authenticated on the machine.
168
184
 
169
- The built-in adapter launches a fresh Claude Code print-mode session for each worker run.
170
- When audit-task review is pending, the provider prompt now asks Claude Code to write structured audit results and then hand back to the bounded worker command so the same wrapper invocation can continue.
185
+ The current implementation can launch a fresh Claude Code print-mode session for each worker run.
171
186
 
172
- Recommended when you want the audit supervisor to delegate bounded tasks into Claude Code without manually driving each step.
187
+ Treat this as a compatibility bridge only, not as the intended default review owner.
173
188
 
174
189
  ### opencode
175
190
 
176
191
  Use when OpenCode is installed and authenticated on the machine.
177
192
 
178
- The built-in adapter launches a fresh `opencode run ...` session for each worker run.
179
- When audit-task review is pending, the provider prompt now asks OpenCode to write structured audit results and then hand back to the bounded worker command so the same wrapper invocation can continue.
193
+ The current implementation can launch a fresh `opencode run ...` session for each worker run.
180
194
 
181
- Recommended when OpenCode is the preferred local agent surface.
195
+ Treat this as a compatibility bridge only, not as the intended default review owner.
182
196
 
183
197
  ### subprocess-template
184
198
 
@@ -195,11 +209,15 @@ Treat this as an advanced backend adapter rather than the default path.
195
209
 
196
210
  ### Claude Code
197
211
 
198
- Use the repo-local `audit-code` wrapper from the target repository root, or set `provider` to `claude-code` in `.audit-artifacts/session-config.json` so the supervisor delegates bounded worker runs into Claude Code.
212
+ Use `/audit-code` in the active conversation as the primary path.
213
+
214
+ Only use the repo-local `audit-code` wrapper with `provider: "claude-code"` in `.audit-artifacts/session-config.json` when you intentionally want backend fallback bridging into Claude Code.
199
215
 
200
216
  ### OpenCode
201
217
 
202
- Use the same repo-local `audit-code` wrapper, or set `provider` to `opencode` so the supervisor delegates bounded worker runs into OpenCode.
218
+ Use `/audit-code` in the active conversation as the primary path.
219
+
220
+ Only use the repo-local `audit-code` wrapper with `provider: "opencode"` when you intentionally want backend fallback bridging into OpenCode.
203
221
 
204
222
  ### VS Code
205
223
 
@@ -249,3 +267,5 @@ For a polished operator experience today:
249
267
  3. use `audit-code` as the repo-local backend fallback
250
268
  4. prefer `local-subprocess` unless you want interactive review to continue automatically through agent tasks
251
269
  5. use `subprocess-template` only when integrating a non-native editor or launcher surface
270
+
271
+ If you intentionally want the backend fallback to bridge semantic review into another process, re-run with an explicit `--provider` flag after configuring the matching section in `.audit-artifacts/session-config.json`.
@@ -36,7 +36,24 @@ That means the current release is suitable for a controlled alpha or beta skill-
36
36
 
37
37
  ## Near-term priorities
38
38
 
39
- ### 1. Verify the shipped host integrations end to end
39
+ ### 1. Realign review dispatch with the conversation-owned workflow
40
+
41
+ The highest-priority product refactor is to move semantic-review ownership back to the active conversation agent and to replace the current unit-first review fan-out with non-overlapping lens-aware review blocks.
42
+
43
+ Near-term work should focus on:
44
+
45
+ - making the active conversation agent the default owner of semantic review
46
+ - keeping `agent_task_batch_size` at one review block per task
47
+ - treating backend provider adapters as compatibility bridges rather than the default review owner
48
+ - replacing the current unit-first task planner with a non-overlapping lens-block planner
49
+ - deleting the stale audit state and rerunning the audit only after that refactor lands
50
+
51
+ The current handoff for this work is:
52
+
53
+ - `docs/workflow-refactor-brief.md`
54
+ - `docs/remediation-baseline.md`
55
+
56
+ ### 2. Verify the shipped host integrations end to end
40
57
 
41
58
  The biggest remaining gap is not raw feature presence anymore. It is host-by-host proof that the generated assets work in the actual products they target.
42
59
 
@@ -48,7 +65,7 @@ Near-term work should focus on:
48
65
  - validating the VS Code prompt, agent, and `.vscode/mcp.json` flow inside a real workspace
49
66
  - validating that the Antigravity planning-mode guidance is accurate and does not over-promise a native saved-workflow surface
50
67
 
51
- ### 2. Close the remaining host-native UX gaps
68
+ ### 3. Close the remaining host-native UX gaps
52
69
 
53
70
  The product goal is still conversational first, not fallback-CLI first, and some shipped surfaces are still guidance-heavy rather than truly native.
54
71
 
@@ -59,7 +76,7 @@ Near-term work should focus on:
59
76
  - deciding whether OpenCode and VS Code need any smaller UX refinements after smoke-testing, rather than assuming the first generated surfaces are final
60
77
  - keeping Antigravity framed as a workflow-and-artifacts host until Google documents a stable project-local config surface
61
78
 
62
- ### 3. Polish continuation through assisted review
79
+ ### 4. Polish continuation through assisted review
63
80
 
64
81
  The repo-local backend fallback still intentionally stops in blocked state under `local-subprocess`, but configured provider bridges can now continue the audit-task review phase automatically.
65
82
 
@@ -69,7 +86,7 @@ Near-term work should focus on:
69
86
  - less operator guesswork when a configured provider fails to return usable results
70
87
  - stronger host-specific guidance for provider-assisted bridges
71
88
 
72
- ### 4. Harden publish and release operations
89
+ ### 5. Harden publish and release operations
73
90
 
74
91
  The packaged install story is in place, but release operations still need finishing work.
75
92
 
package/docs/packaging.md CHANGED
@@ -31,6 +31,18 @@ npm install
31
31
  npm run verify:release
32
32
  ```
33
33
 
34
+ For live child-process output during packaged smoke debugging:
35
+
36
+ ```bash
37
+ AUDIT_CODE_VERBOSE=1 npm run smoke:packaged-audit-code
38
+ ```
39
+
40
+ For the linked-install variant:
41
+
42
+ ```bash
43
+ AUDIT_CODE_VERBOSE=1 npm run smoke:linked-audit-code
44
+ ```
45
+
34
46
  That script:
35
47
 
36
48
  - creates a tarball with `npm pack`
@@ -41,6 +53,8 @@ That script:
41
53
  - invokes the installed `audit-code` binary from `node_modules/.bin`
42
54
  - validates emitted JSON against `schemas/audit-code-v1alpha1.schema.json`
43
55
  - exercises the same evidence-import and completion flow used by the linked-command smoke test
56
+ - strips inherited `npm_config_*`, `NODE_AUTH_TOKEN`, and `NPM_TOKEN` values before nested npm operations so `npm publish --dry-run` does not accidentally suppress tarball generation or leak publish credentials into the smoke environment
57
+ - emits step-by-step progress plus a final success summary so CI logs show where the run failed or completed across both the linked and packaged smoke paths
44
58
 
45
59
  ## CI
46
60
 
@@ -78,11 +78,32 @@ For repo-local backend usage:
78
78
  - `provider: "auto"` is the explicit opt-in mode for best-effort routing across configured or detected backends
79
79
  - explicit provider names should remain available for operators who want a specific backend
80
80
 
81
+ ## Semantic review ownership
82
+
83
+ The intended semantic-review owner is the active conversation agent.
84
+
85
+ That means:
86
+
87
+ - `/audit-code` should hand review work to the current conversation or host agent session first
88
+ - if that active agent can delegate to subagents in parallel, that fan-out belongs to the host runtime rather than to repo-local backend defaults
89
+ - backend provider adapters are compatibility bridges for fallback CLI usage, not the default owner of review work
90
+ - session-config should not be the normal mechanism for redirecting semantic review into a second external LLM CLI
91
+
92
+ ## Task-planning rule
93
+
94
+ The intended review planner should:
95
+
96
+ - determine which files require which lenses
97
+ - partition unresolved review into non-overlapping review blocks
98
+ - prefer lens-homogeneous blocks when practical
99
+ - keep the default dispatch granularity to one review block per task
100
+
81
101
  ## Default context & model rules
82
102
 
83
103
  1. By default, the current ChatGPT project conversation and its files should be treated as the primary context.
84
104
  2. The user should not need to supply `--root`, provider names, or backend-specific settings in normal usage.
85
105
  3. For backend CLI delegation, let the chosen provider own its own model-selection behavior unless explicitly configured.
106
+ 4. Backend fallback settings should not cap the active conversation agent's own subagent parallelism model.
86
107
 
87
108
  ## Development rule
88
109
 
@@ -92,6 +113,7 @@ Future development should optimize for the native skill UX first:
92
113
  - packaged installs should help users reach the prompt asset for that conversation route
93
114
  - the CLI and supervisor are implementation details and fallback harnesses
94
115
  - provider adapters are backend internals, not the primary product concept
116
+ - semantic-review dispatch belongs to the active conversation agent, not to an externally spawned fallback provider by default
95
117
  - docs, tests, and examples should present the skill-first flow before any CLI flow
96
118
 
97
119
  If documentation or implementation details conflict, prefer the skill-first contract above over the CLI-first backend shape.
@@ -20,6 +20,8 @@ Anything below `dist/index.js` remains a backend or development interface rather
20
20
  ### Node and package runtime
21
21
 
22
22
  - Node `>=20` is the minimum supported runtime from `package.json`
23
+ - GitHub Actions currently exercises the test suite on Node `20` and Node `22`
24
+ - release and publish automation stays pinned to Node `22.14.0`
23
25
  - packaged installs must include:
24
26
  - `audit-code`
25
27
  - `audit-code-wrapper-lib.mjs`
package/docs/releasing.md CHANGED
@@ -13,7 +13,9 @@ That workflow already:
13
13
  - pins Node `22.14.0`
14
14
  - upgrades npm to `>=11.5.1`
15
15
  - runs `npm run verify:release` before any publish attempt
16
+ - previews the packed tarball with `npm pack --dry-run` before any live publish attempt
16
17
  - defaults semver prerelease versions to the `next` dist-tag unless manual dispatch overrides it
18
+ - uploads npm debug logs when a publish-path step fails
17
19
 
18
20
  ## One-time npm setup
19
21
 
@@ -39,6 +41,12 @@ npm publish --dry-run
39
41
 
40
42
  The local dry run is still valuable even though the real publish happens in GitHub Actions. It proves the packed artifact, lifecycle hooks, and publish metadata before you spend a release on a broken workflow run.
41
43
 
44
+ ## Supported Node lines
45
+
46
+ Routine CI currently exercises the repository on Node `20` and Node `22`.
47
+
48
+ Release and publish workflows stay pinned to Node `22.14.0` because that is the line this repository uses for npm Trusted Publishing and release smoke verification.
49
+
42
50
  ## GitHub release paths
43
51
 
44
52
  ### Manual dry run
@@ -68,6 +76,15 @@ This is the safest path for the first trusted-publishing release because it lets
68
76
 
69
77
  After the manual path is proven, publishing a GitHub Release will trigger the same workflow and publish the package.
70
78
 
79
+ ## Workflow troubleshooting
80
+
81
+ If a GitHub Actions run fails:
82
+
83
+ 1. download the uploaded `*-npm-logs` artifact from the failed run
84
+ 2. rerun `npm ci` and `npm run verify:release` locally from the same commit
85
+ 3. for publish failures, rerun `publish-package.yml` with `dry_run=true` before retrying a live publish
86
+ 4. if publish still fails, confirm npm Trusted Publishing is still configured for `publish-package.yml`
87
+
71
88
  ## Post-publish checks
72
89
 
73
90
  After a live publish, verify the result from a clean shell: