agentera 3.0.0-dev.8 → 3.0.0-dev.81

package/dist/analytics/extractCorpus/corpus.js

@@ -1,5 +1,5 @@
 import fs from "node:fs";
-import { ADAPTER_VERSION, FAMILIES, isoNow, runtimeStatus, discoverRuntimeStore, COPILOT_SPARSE_REMEDIATION, } from "./core.js";
+import { ADAPTER_VERSION, FAMILIES, isoNow, runtimeStatus, discoverRuntimeStore, COPILOT_SPARSE_REMEDIATION, assessProvenance, } from "./core.js";
 import { isPlainObject } from "./core.js";
 import { extractInstructionDocuments, extractProjectConfigSignals } from "./filesystemSources.js";
 import { extractCodexSessions, extractClaudeProjectSessions } from "./jsonlSessions.js";
@@ -10,8 +10,12 @@ import { resolvePath } from "../../core/paths.js";
 import { applyTruncationToStatus } from "./sqliteCaps.js";
 export class ExtractionNotImplementedError extends Error {
 }
-function extractRuntimeStore(runtime, storePath, errors, extractor, sqliteCaps) {
-    const discovery = discoverRuntimeStore(runtime, storePath);
+function extractRuntimeStore(runtime, storePath, errors, extractor, sqliteCaps, provenance = {
+    sourceProduct: runtime ?? "unknown",
+    sourceClass: "active_runtime",
+    activeRuntime: true,
+}) {
+    const discovery = discoverRuntimeStore(runtime, storePath, provenance);
     if (discovery.status !== "available")
         return [[], discovery];
     const errorStart = errors.length;
@@ -23,17 +27,34 @@ function extractRuntimeStore(runtime, storePath, errors, extractor, sqliteCaps)
     catch (exc) {
         const fc = (discovery.file_count ?? null); // cast: discovery payload IO boundary
         if (exc instanceof ExtractionNotImplementedError) {
-            return [[], runtimeStatus(runtime, { status: "degraded", reason: "extractor_unimplemented", storePath, fileCount: fc, recordCount: 0, errorCount: 0 })];
+            return [[], runtimeStatus(runtime, { status: "degraded", reason: "extractor_unimplemented", storePath, fileCount: fc, recordCount: 0, errorCount: 0, ...provenance })];
         }
         if (exc instanceof PermissionDeniedError) {
-            return [[], runtimeStatus(runtime, { status: "degraded", reason: "store_locked", storePath, fileCount: fc })];
+            return [[], runtimeStatus(runtime, { status: "degraded", reason: "store_locked", storePath, fileCount: fc, ...provenance })];
         }
-        return [[], runtimeStatus(runtime, { status: "degraded", reason: "store_unreadable", storePath, fileCount: fc })];
+        return [[], runtimeStatus(runtime, { status: "degraded", reason: "store_unreadable", storePath, fileCount: fc, ...provenance })];
     }
     const fc = (discovery.file_count ?? null); // cast: discovery payload IO boundary
     const errorCount = errors.length - errorStart;
+    const recordProvenance = assessProvenance(records);
+    if (!recordProvenance.complete) {
+        return [
+            records,
+            runtimeStatus(runtime, {
+                status: "degraded",
+                reason: "provenance_missing",
+                storePath,
+                fileCount: fc,
+                recordCount: records.length,
+                errorCount,
+                provenanceMissingFields: recordProvenance.missingFields,
+                provenanceMissingRecords: recordProvenance.missingRecords,
+                ...provenance,
+            }),
+        ];
+    }
     if (errorCount) {
-        return [records, runtimeStatus(runtime, { status: "degraded", reason: "schema_divergent", storePath, fileCount: fc, recordCount: records.length, errorCount })];
+        return [records, runtimeStatus(runtime, { status: "degraded", reason: "schema_divergent", storePath, fileCount: fc, recordCount: records.length, errorCount, ...provenance })];
     }
     if (records.length === 0) {
         return [records, runtimeStatus(runtime, {
@@ -42,12 +63,13 @@ function extractRuntimeStore(runtime, storePath, errors, extractor, sqliteCaps)
                 storePath,
                 fileCount: fc,
                 recordCount: 0,
-                remediationLabels: runtime === "github-copilot" ? [COPILOT_SPARSE_REMEDIATION] : null,
+                remediationLabels: runtime === "copilot" ? [COPILOT_SPARSE_REMEDIATION] : null,
+                ...provenance,
             })];
     }
     return [
         records,
-        applyTruncationToStatus(runtimeStatus(runtime, { status: "ok", reason: "records_extracted", storePath, fileCount: fc, recordCount: records.length, errorCount: 0 }), ctx.truncation),
+        applyTruncationToStatus(runtimeStatus(runtime, { status: "ok", reason: "records_extracted", storePath, fileCount: fc, recordCount: records.length, errorCount: 0, ...provenance }), ctx.truncation),
     ];
 }
 export function dedupeRecords(records) {
@@ -90,7 +112,9 @@ export function buildMetadata(records, errors, runtimeStatuses, coverage) {
         if (count === 0)
             families[family].error = "no records extracted for this family";
     }
-    const runtimes = Array.from(new Set(records.filter((i) => i.runtime).map((i) => String(i.runtime)))).sort();
+    const runtimes = Array.from(new Set(records.filter((item) => item.active_runtime === true && item.runtime).map((item) => String(item.runtime)))).sort();
+    const sourceClasses = Array.from(new Set(records.map((item) => String(item.source_class)))).sort();
+    const sourceProducts = Array.from(new Set(records.map((item) => String(item.source_product)))).sort();
     const coverageFields = coverage ?? {
         available_runtimes: [],
         selected_runtimes: [],
@@ -107,6 +131,8 @@ export function buildMetadata(records, errors, runtimeStatuses, coverage) {
         available_but_not_selected: coverageFields.available_but_not_selected,
         total_records: records.length,
         errors,
+        source_classes: sourceClasses,
+        source_products: sourceProducts,
     };
 }
 export function buildCorpus(opts) {
@@ -123,23 +149,36 @@ export function buildCorpus(opts) {
     records.push(...extractProjectConfigSignals(normalizedRoots, errors));
     const runtimeStatuses = [];
     const runtimes = [
-        ["codex", opts.codexSessionsDir, extractCodexSessions],
-        ["claude-code", opts.claudeProjectsDir, extractClaudeProjectSessions],
-        ["cursor", opts.cursorProjectsDir ?? null, (sp, err) => extractCursorSessions(sp, err, normalizedRoots)],
+        ["codex", "codex", "active_runtime", true, opts.codexSessionsDir, extractCodexSessions, undefined],
+        ["cursor", "cursor", "active_runtime", true, opts.cursorProjectsDir ?? null, (sp, err) => extractCursorSessions(sp, err, normalizedRoots), undefined],
         [
+            "cursor",
             "cursor-agent",
+            "active_runtime",
+            true,
             opts.cursorChatsDir ?? null,
             (sp, err) => extractCursorAgentSessions(sp, err, normalizedRoots, opts.cursorProjectsDir ?? null),
+            "store.db",
         ],
-        ["opencode", opts.opencodeConversationsDir ?? null, extractOpencodeSessions],
-        ["github-copilot", opts.copilotConversationsDir ?? null, extractCopilotSessions],
+        ["opencode", "opencode", "active_runtime", true, opts.opencodeConversationsDir ?? null, extractOpencodeSessions, undefined],
+        ["copilot", "github-copilot", "active_runtime", true, opts.copilotConversationsDir ?? null, extractCopilotSessions, undefined],
     ];
-    for (const [runtime, storePath, extractor] of runtimes) {
-        const [runtimeRecords, status] = extractRuntimeStore(runtime, storePath, errors, extractor, opts.sqliteCaps);
+    if (opts.claudeProjectsDir !== null) {
+        runtimes.push([
+            null,
+            "claude-code",
+            "historical_import",
+            false,
+            opts.claudeProjectsDir,
+            extractClaudeProjectSessions,
+            "*.jsonl",
+        ]);
+    }
+    for (const [runtime, sourceProduct, sourceClass, activeRuntime, storePath, extractor, pattern] of runtimes) {
+        const [runtimeRecords, status] = extractRuntimeStore(runtime, storePath, errors, extractor, opts.sqliteCaps, { sourceProduct, sourceClass, activeRuntime, pattern });
         records.push(...runtimeRecords);
         runtimeStatuses.push(status);
     }
     const deduped = dedupeRecords(records);
     return { metadata: buildMetadata(deduped, errors, runtimeStatuses, opts.coverage), records: deduped };
 }
-//# sourceMappingURL=corpus.js.map

package/dist/analytics/extractCorpus/coverageAudit.js

@@ -14,36 +14,27 @@ export function corpusEnvelopeCoverage(audit) {
         })),
     };
 }
-export function resolveRuntimeStoreConfigs(args, env = process.env, platform = process.platform) {
+export function resolveRuntimeStoreConfigs(args, env = process.env, _platform = process.platform) {
     const configs = [
-        ["codex", args.codexSessionsDir, !args.noCodex, args.noCodex ? "disabled_by_flag" : null],
-        ["claude-code", args.claudeProjectsDir, !args.noClaude, args.noClaude ? "disabled_by_flag" : null],
-        ["cursor", args.cursorProjectsDir || resolveCursorProjectsPath(env), !args.noCursor, args.noCursor ? "disabled_by_flag" : null],
-        [
-            "cursor-agent",
-            args.cursorChatsDir || resolveCursorChatsPath(env),
-            !args.noCursor,
-            args.noCursor ? "disabled_by_flag" : null,
-        ],
-        [
-            "opencode",
-            args.opencodeConversationsDir || resolveOpencodeDbPath(env),
-            !args.noOpencode,
-            args.noOpencode ? "disabled_by_flag" : null,
-        ],
-        [
-            "github-copilot",
-            args.copilotConversationsDir || resolveCopilotStorePath(env),
-            !args.noCopilot,
-            args.noCopilot ? "disabled_by_flag" : null,
-        ],
+        { runtime: "codex", sourceProduct: "codex", sourceClass: "active_runtime", activeRuntime: true, storePath: args.codexSessionsDir, selected: !args.noCodex, skipReason: args.noCodex ? "disabled_by_flag" : null },
+        { runtime: "cursor", sourceProduct: "cursor", sourceClass: "active_runtime", activeRuntime: true, storePath: args.cursorProjectsDir || resolveCursorProjectsPath(env), selected: !args.noCursor, skipReason: args.noCursor ? "disabled_by_flag" : null },
+        { runtime: "cursor", sourceProduct: "cursor-agent", sourceClass: "active_runtime", activeRuntime: true, pattern: "store.db", storePath: args.cursorChatsDir || resolveCursorChatsPath(env), selected: !args.noCursor, skipReason: args.noCursor ? "disabled_by_flag" : null },
+        { runtime: "opencode", sourceProduct: "opencode", sourceClass: "active_runtime", activeRuntime: true, storePath: args.opencodeConversationsDir || resolveOpencodeDbPath(env), selected: !args.noOpencode, skipReason: args.noOpencode ? "disabled_by_flag" : null },
+        { runtime: "copilot", sourceProduct: "github-copilot", sourceClass: "active_runtime", activeRuntime: true, storePath: args.copilotConversationsDir || resolveCopilotStorePath(env), selected: !args.noCopilot, skipReason: args.noCopilot ? "disabled_by_flag" : null },
     ];
-    return configs.map(([runtime, storePath, selected, skipReason]) => ({
-        runtime,
-        storePath: storePath ?? null,
-        selected,
-        skipReason,
-    }));
+    if ((args.importSources ?? []).includes("claude")) {
+        configs.push({
+            runtime: null,
+            sourceProduct: "claude-code",
+            sourceClass: "historical_import",
+            activeRuntime: false,
+            pattern: "*.jsonl",
+            storePath: args.claudeProjectsDir,
+            selected: true,
+            skipReason: null,
+        });
+    }
+    return configs.map((config) => ({ ...config, storePath: config.storePath ?? null }));
 }
 function trackEarliest(current, candidate) {
     if (!candidate)
@@ -78,10 +69,10 @@ function probeJsonlTimestamps(storePath) {
     }
     return { earliest, latest };
 }
-function probeSqliteTimestamps(storePath, runtime) {
-    const dbPaths = runtime === "opencode"
+function probeSqliteTimestamps(storePath, sourceProduct) {
+    const dbPaths = sourceProduct === "opencode"
         ? [storePath]
-        : runtime === "github-copilot"
+        : sourceProduct === "github-copilot"
             ? isFilePath(storePath)
                 ? [storePath]
                 : rglob(storePath, "session-store.db")
@@ -102,7 +93,7 @@ function probeSqliteTimestamps(storePath, runtime) {
             continue;
         }
         try {
-            if (runtime === "opencode") {
+            if (sourceProduct === "opencode") {
                 const cols = tableColumns(conn, "session");
                 const timeCol = firstColumn(cols, ["time_created", "time", "timestamp", "created_at", "createdAt"]);
                 if (timeCol) {
@@ -117,7 +108,7 @@ function probeSqliteTimestamps(storePath, runtime) {
                     }
                 }
             }
-            else if (runtime === "github-copilot") {
+            else if (sourceProduct === "github-copilot") {
                 const sessionCols = tableColumns(conn, "sessions");
                 const turnCols = tableColumns(conn, "turns");
                 const sessionTime = firstColumn(sessionCols, ["time", "timestamp", "created_at", "createdAt"]);
@@ -135,7 +126,7 @@ function probeSqliteTimestamps(storePath, runtime) {
                     }
                 }
             }
-            else if (runtime === "cursor-agent") {
+            else if (sourceProduct === "cursor-agent") {
                 const rows = conn.prepare("SELECT data FROM blobs ORDER BY id").all();
                 for (const row of rows) {
                     const payload = row.data;
@@ -173,26 +164,31 @@ function probeSqliteTimestamps(storePath, runtime) {
     }
     return { earliest, latest };
 }
-function probeRuntimeTimestamps(runtime, storePath) {
-    if (runtime === "opencode" || runtime === "github-copilot" || runtime === "cursor-agent") {
-        return probeSqliteTimestamps(storePath, runtime);
+function probeRuntimeTimestamps(sourceProduct, storePath) {
+    if (sourceProduct === "opencode" || sourceProduct === "github-copilot" || sourceProduct === "cursor-agent") {
+        return probeSqliteTimestamps(storePath, sourceProduct);
     }
     if (isDir(storePath))
         return probeJsonlTimestamps(storePath);
     if (isFilePath(storePath))
-        return probeSqliteTimestamps(storePath, runtime);
+        return probeSqliteTimestamps(storePath, sourceProduct);
     return { earliest: null, latest: null };
 }
 export function runCoverageAudit(args, env = process.env, platform = process.platform, acceptCoverageGap = false) {
     const runtimes = [];
     const skippedAvailable = [];
     for (const config of resolveRuntimeStoreConfigs(args, env, platform)) {
-        const discovery = discoverRuntimeStore(config.runtime, config.storePath);
+        const discovery = discoverRuntimeStore(config.runtime, config.storePath, {
+            sourceProduct: config.sourceProduct,
+            sourceClass: config.sourceClass,
+            activeRuntime: config.activeRuntime,
+            pattern: config.pattern,
+        });
         const available = discovery.status === "available";
         let earliest = null;
         let latest = null;
         if (available && config.storePath) {
-            const bounds = probeRuntimeTimestamps(config.runtime, config.storePath);
+            const bounds = probeRuntimeTimestamps(config.sourceProduct, config.storePath);
             earliest = bounds.earliest;
             latest = bounds.latest;
         }
@@ -202,6 +198,9 @@ export function runCoverageAudit(args, env = process.env, platform = process.pla
         }
         runtimes.push({
             runtime: config.runtime,
+            source_product: config.sourceProduct,
+            source_class: config.sourceClass,
+            active_runtime: config.activeRuntime,
             store_path: config.storePath,
             selected: config.selected,
             discovery_status: String(discovery.status),
@@ -212,8 +211,8 @@ export function runCoverageAudit(args, env = process.env, platform = process.pla
             skip_reason: skipReason,
         });
     }
-    const availableRuntimes = runtimes.filter((r) => r.available).map((r) => r.runtime);
-    const selectedRuntimes = runtimes.filter((r) => r.selected).map((r) => r.runtime);
+    const availableRuntimes = [...new Set(runtimes.filter((r) => r.active_runtime && r.available).map((r) => r.runtime))];
+    const selectedRuntimes = [...new Set(runtimes.filter((r) => r.active_runtime && r.selected).map((r) => r.runtime))];
     const coverageGapFlagged = skippedAvailable.length > 0 && !acceptCoverageGap;
     return {
         runtimes,
@@ -232,10 +231,11 @@ export function formatCoverageSummaryText(audit) {
             const span = entry.earliest_session && entry.latest_session
                 ? `${entry.earliest_session} .. ${entry.latest_session}`
                 : "timestamps unavailable";
-            lines.push(`  ${entry.runtime}: available store=${store} sessions=${span} selected=${entry.selected ? "yes" : "no"}`);
+            const label = entry.active_runtime ? entry.runtime : `${entry.source_product} [historical import]`;
+            lines.push(`  ${label}: available store=${store} sessions=${span} selected=${entry.selected ? "yes" : "no"}`);
         }
         else {
-            lines.push(`  ${entry.runtime}: ${entry.discovery_status} (${entry.discovery_reason}) store=${store} selected=${entry.selected ? "yes" : "no"}`);
+            lines.push(`  ${entry.active_runtime ? entry.runtime : `${entry.source_product} [historical import]`}: ${entry.discovery_status} (${entry.discovery_reason}) store=${store} selected=${entry.selected ? "yes" : "no"}`);
         }
     }
     if (audit.skipped_available.length > 0) {
@@ -258,4 +258,3 @@ export function formatCoverageSummaryText(audit) {
     }
     return lines.join("\n");
 }
-//# sourceMappingURL=coverageAudit.js.map

package/dist/analytics/extractCorpus/cursorSessions.js

@@ -4,7 +4,7 @@ import path from "node:path";
 import crypto from "node:crypto";
 import { execFileSync } from "node:child_process";
 import { expanduser, resolvePath } from "../../core/paths.js";
-import { MAX_TOOL_ARG_TEXT, eventTimestamp, isFilePath, isoFromMtime, iterJsonl, record, signalType, textFromContent, toolCallRecordFromItem, } from "./core.js";
+import { authorClassForRole, MAX_TOOL_ARG_TEXT, eventTimestamp, isFilePath, isoFromMtime, iterJsonl, record, signalType, textFromContent, transportProvenance, toolCallRecordFromItem, } from "./core.js";
 import { isPlainObject, rglob, isDir } from "./core.js";
 import { openSqlite, } from "./sqliteSessions.js";
 function pathStem(p) {
@@ -153,7 +153,7 @@ export function extractCursorSessions(projectsDir, errors, projectRoots = null)
         for (const event of iterJsonl(p, errors)) {
             index += 1;
             const role = event.role;
-            if (role !== "user" && role !== "assistant")
+            if (typeof role !== "string" || role.length === 0)
                 continue;
             let contentIndex = 0;
             for (const item of cursorContentItems(event)) {
@@ -180,6 +180,9 @@ export function extractCursorSessions(projectsDir, errors, projectRoots = null)
             if (!content)
                 continue;
             const timestamp = eventTimestamp(event, fallbackTimestamp);
+            const transport = transportProvenance(event, ...items);
+            const originId = transport.originId;
+            const authorClass = authorClassForRole(role, transport.authorClass, transport.originId);
             const data = { actor: role, content };
             if (role === "user") {
                 if (previousAssistant)
@@ -198,6 +201,9 @@ export function extractCursorSessions(projectsDir, errors, projectRoots = null)
                 runtime: "cursor",
                 sourceParts: [resolvePath(p), index, role, content.slice(0, 80)],
                 sessionId,
+                originId,
+                authorClass,
+                content,
                 data,
             }));
             if (role === "user") {
@@ -210,6 +216,9 @@ export function extractCursorSessions(projectsDir, errors, projectRoots = null)
                         runtime: "cursor",
                         sourceParts: [resolvePath(p), index, "history", content.slice(0, 120)],
                         sessionId,
+                        originId,
+                        authorClass,
+                        content,
                         data: { prompt: content, signal_type: sig },
                     }));
                 }
@@ -356,6 +365,8 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
         for (const [blobId, message] of cursorAgentBlobMessages(storeDb, errors)) {
             index += 1;
             const role = message.role;
+            if (typeof role !== "string" || role.length === 0)
+                continue;
             let toolIndex = 0;
             for (const toolItem of cursorAgentToolItems(message)) {
                 toolIndex += 1;
@@ -364,7 +375,8 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
                     event: message,
                     fallbackTimestamp,
                     projectPath,
-                    runtime: "cursor-agent",
+                    runtime: "cursor",
+                    sourceProduct: "cursor-agent",
                     sourcePath: storeDb,
                     index: index * 1000 + toolIndex,
                     sessionId,
@@ -376,12 +388,13 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
                     records.push(tr);
                 }
             }
-            if (role !== "user" && role !== "assistant")
-                continue;
             const content = cursorAgentMessageText(message);
             if (!content)
                 continue;
             const timestamp = eventTimestamp(message, fallbackTimestamp);
+            const transport = transportProvenance(message);
+            const originId = transport.originId;
+            const authorClass = authorClassForRole(role, transport.authorClass, transport.originId);
             const data = { actor: role, content };
             if (role === "user") {
                 if (previousAssistant)
@@ -397,9 +410,13 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
                 sourceKind: "conversation_turn",
                 timestamp,
                 projectPath,
-                runtime: "cursor-agent",
+                runtime: "cursor",
+                sourceProduct: "cursor-agent",
                 sourceParts: [resolvePath(storeDb), blobId, role, content.slice(0, 80)],
                 sessionId,
+                originId,
+                authorClass,
+                content,
                 data,
             }));
             if (role === "user") {
@@ -409,9 +426,13 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
                         sourceKind: "history_prompt",
                         timestamp,
                         projectPath,
-                        runtime: "cursor-agent",
+                        runtime: "cursor",
+                        sourceProduct: "cursor-agent",
                         sourceParts: [resolvePath(storeDb), blobId, "history", content.slice(0, 120)],
                         sessionId,
+                        originId,
+                        authorClass,
+                        content,
                         data: { prompt: content, signal_type: sig },
                     }));
                 }
@@ -420,4 +441,3 @@ export function extractCursorAgentSessions(chatsDir, errors, projectRoots = null
     }
     return records;
 }
-//# sourceMappingURL=cursorSessions.js.map