@tonyclaw/agent-inspector 2.0.29 → 2.0.31

package/.output/server/_tanstack-start-manifest_v-BVM4AUlx.mjs

@@ -0,0 +1,4 @@
+const tsrStartManifest = () => ({ "routes": { "__root__": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/__root.tsx", "children": ["/", "/api/config", "/api/groups", "/api/health", "/api/logs", "/api/mcp", "/api/models", "/api/providers", "/api/runs", "/api/sessions", "/proxy/$", "/session/$sessionId", "/api/knowledge/candidates", "/api/knowledge/project-context", "/api/knowledge/search", "/api/knowledge/sessions/$sessionId/candidates"], "preloads": ["/assets/main-B1kdhY98.js"], "assets": [] }, "/": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/index.tsx", "assets": [], "preloads": ["/assets/index-Kxptlkpg.js", "/assets/ProxyViewerContainer-BxRaXhKR.js"] }, "/api/config": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/config.ts", "children": ["/api/config/paths"] }, "/api/groups": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/groups.ts", "children": ["/api/groups/$groupId"] }, "/api/health": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/health.ts" }, "/api/logs": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/logs.ts", "children": ["/api/logs/$id", "/api/logs/stream"] }, "/api/mcp": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/mcp.ts" }, "/api/models": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/models.ts" }, "/api/providers": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.ts", "children": ["/api/providers/$providerId", "/api/providers/export", "/api/providers/import", "/api/providers/scan"] }, "/api/runs": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/runs.ts", "children": ["/api/runs/$runId"] }, "/api/sessions": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/sessions.ts" }, "/proxy/$": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/proxy/$.ts" }, "/session/$sessionId": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/session/$sessionId.tsx", "assets": [], "preloads": ["/assets/_sessionId-BhNTuZ31.js", "/assets/ProxyViewerContainer-BxRaXhKR.js"] }, "/api/config/paths": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/config.paths.ts" }, "/api/groups/$groupId": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/groups.$groupId.ts", "children": ["/api/groups/$groupId/evidence", "/api/groups/$groupId/sessions"] }, "/api/knowledge/candidates": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.candidates.ts", "children": ["/api/knowledge/candidates/$candidateId"] }, "/api/knowledge/project-context": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.project-context.ts" }, "/api/knowledge/search": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.search.ts" }, "/api/logs/$id": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/logs.$id.ts", "children": ["/api/logs/$id/chunks", "/api/logs/$id/replay"] }, "/api/logs/stream": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/logs.stream.ts" }, "/api/providers/$providerId": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.$providerId.ts", "children": ["/api/providers/$providerId/model-metadata", "/api/providers/$providerId/test"] }, "/api/providers/export": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.export.ts" }, "/api/providers/import": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.import.ts" }, "/api/providers/scan": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.scan.ts" }, "/api/runs/$runId": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/runs.$runId.ts", "children": ["/api/runs/$runId/evidence"] }, "/api/groups/$groupId/evidence": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/groups.$groupId.evidence.ts" }, "/api/groups/$groupId/sessions": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/groups.$groupId.sessions.ts" }, "/api/knowledge/candidates/$candidateId": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.candidates.$candidateId.ts", "children": ["/api/knowledge/candidates/$candidateId/promote"] }, "/api/logs/$id/chunks": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/logs.$id.chunks.ts" }, "/api/logs/$id/replay": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/logs.$id.replay.ts" }, "/api/providers/$providerId/model-metadata": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.$providerId.model-metadata.ts" }, "/api/providers/$providerId/test": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.$providerId.test.ts", "children": ["/api/providers/$providerId/test/log"] }, "/api/runs/$runId/evidence": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/runs.$runId.evidence.ts" }, "/api/knowledge/candidates/$candidateId/promote": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.candidates.$candidateId.promote.ts" }, "/api/knowledge/sessions/$sessionId/candidates": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/knowledge.sessions.$sessionId.candidates.ts" }, "/api/providers/$providerId/test/log": { "filePath": "C:/Users/claw/workspace/agent-inspector/src/routes/api/providers.$providerId.test.log.ts" } }, "clientEntry": "/assets/main-B1kdhY98.js" });
+export {
+  tsrStartManifest
+};

package/.output/server/index.mjs

@@ -35,103 +35,103 @@ const headers = ((m) => function headersRouteRule(event) {
   }
 });
 const assets = {
+  "/assets/index-Kxptlkpg.js": {
+    "type": "text/javascript; charset=utf-8",
+    "etag": '"74-zVU85yxwksDEAlmGvjnynbRJlsg"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 116,
+    "path": "../public/assets/index-Kxptlkpg.js"
+  },
+  "/assets/index-Byk60-jA.css": {
+    "type": "text/css; charset=utf-8",
+    "etag": '"1ab1a-tPrFGLVNN1Xk1g4iKBR5XFS0VgE"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 109338,
+    "path": "../public/assets/index-Byk60-jA.css"
+  },
   "/assets/alibaba-TTwafVwX.svg": {
     "type": "image/svg+xml",
     "etag": '"171b-6dyV5K8QjiaY35sN9qNprh9zDIs"',
-    "mtime": "2026-06-30T09:01:48.556Z",
+    "mtime": "2026-07-01T02:23:48.448Z",
     "size": 5915,
     "path": "../public/assets/alibaba-TTwafVwX.svg"
   },
-  "/assets/index-BhQGSdhG.js": {
+  "/assets/CompareDrawer-tIUf2EJm.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"74-1e7gU3IOVaEjblxPQfgH61GRjlw"',
-    "mtime": "2026-06-30T09:01:48.556Z",
-    "size": 116,
-    "path": "../public/assets/index-BhQGSdhG.js"
+    "etag": '"49fe-O2b82jiqVijsJaEWXyXHfD26ujs"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 18942,
+    "path": "../public/assets/CompareDrawer-tIUf2EJm.js"
   },
-  "/assets/CompareDrawer-DcqxlgSQ.js": {
+  "/assets/ReplayDialog-D6v1xcuC.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"49fe-+D2I+/9sd7EYksKARn3HVY2eJf0"',
-    "mtime": "2026-06-30T09:01:48.556Z",
-    "size": 18942,
-    "path": "../public/assets/CompareDrawer-DcqxlgSQ.js"
+    "etag": '"318c-clYQ+hDlXIV8AGEUD0khiYhAzmM"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 12684,
+    "path": "../public/assets/ReplayDialog-D6v1xcuC.js"
+  },
+  "/assets/main-B1kdhY98.js": {
+    "type": "text/javascript; charset=utf-8",
+    "etag": '"51726-il8QGn3n9HZk3cnlzLAvqWtvO8Y"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 333606,
+    "path": "../public/assets/main-B1kdhY98.js"
   },
   "/assets/minimax-BPMzvuL-.jpeg": {
     "type": "image/jpeg",
     "etag": '"1b06-IwivU89ko5UTMUM1/t7hn4sQK9A"',
-    "mtime": "2026-06-30T09:01:48.556Z",
+    "mtime": "2026-07-01T02:23:48.448Z",
     "size": 6918,
     "path": "../public/assets/minimax-BPMzvuL-.jpeg"
   },
-  "/assets/index-CwlHPmgL.css": {
-    "type": "text/css; charset=utf-8",
-    "etag": '"1a3f0-E2f0kOdSBi1RhjoK38+CSij/apg"',
-    "mtime": "2026-06-30T09:01:48.556Z",
-    "size": 107504,
-    "path": "../public/assets/index-CwlHPmgL.css"
-  },
-  "/assets/ReplayDialog-CBcnPgx1.js": {
+  "/assets/RequestAnatomy-D6PniTlx.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"318c-Wq35vSCO67O3pnw6riReoXT9gOg"',
-    "mtime": "2026-06-30T09:01:48.557Z",
-    "size": 12684,
-    "path": "../public/assets/ReplayDialog-CBcnPgx1.js"
-  },
-  "/assets/RequestAnatomy-DSgSqCYt.js": {
-    "type": "text/javascript; charset=utf-8",
-    "etag": '"4485-rBai6wnmYHf662t3omQFALcwHSw"',
-    "mtime": "2026-06-30T09:01:48.557Z",
+    "etag": '"4485-cUU3kN9eHR4o5pbZQbcY4OqCB50"',
+    "mtime": "2026-07-01T02:23:48.450Z",
     "size": 17541,
-    "path": "../public/assets/RequestAnatomy-DSgSqCYt.js"
+    "path": "../public/assets/RequestAnatomy-D6PniTlx.js"
   },
-  "/assets/ResponseView-CjqzBSoF.js": {
+  "/assets/StreamingChunkSequence-D-Rbfnxh.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"30d0-13nPwe9jIZSSl43vTEhXlswXwn8"',
-    "mtime": "2026-06-30T09:01:48.557Z",
-    "size": 12496,
-    "path": "../public/assets/ResponseView-CjqzBSoF.js"
-  },
-  "/assets/StreamingChunkSequence-DdBBKolI.js": {
-    "type": "text/javascript; charset=utf-8",
-    "etag": '"d57-9H1QMwesXGWHkYNTPJ/9zDoslX0"',
-    "mtime": "2026-06-30T09:01:48.557Z",
+    "etag": '"d57-a29gzePmbYwiutnSghD7IFULcvE"',
+    "mtime": "2026-07-01T02:23:48.450Z",
     "size": 3415,
-    "path": "../public/assets/StreamingChunkSequence-DdBBKolI.js"
-  },
-  "/assets/_sessionId-wctOoGh_.js": {
-    "type": "text/javascript; charset=utf-8",
-    "etag": '"d2-fiIvvYyX9jF4fd1W7HpowwQ2Zj4"',
-    "mtime": "2026-06-30T09:01:48.556Z",
-    "size": 210,
-    "path": "../public/assets/_sessionId-wctOoGh_.js"
+    "path": "../public/assets/StreamingChunkSequence-D-Rbfnxh.js"
   },
-  "/assets/main-B6OLZCp9.js": {
+  "/assets/ResponseView-Sx8PjuvU.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"51726-MaBaigWv6OLiiYg6fusFyN758Hw"',
-    "mtime": "2026-06-30T09:01:48.557Z",
-    "size": 333606,
-    "path": "../public/assets/main-B6OLZCp9.js"
+    "etag": '"30d0-eVm2iIA3EKpWcIYtelIeobfqtiY"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 12496,
+    "path": "../public/assets/ResponseView-Sx8PjuvU.js"
   },
   "/assets/zhipuai-BPNAnxo-.svg": {
     "type": "image/svg+xml",
     "etag": '"2bf8-hNaLCTi89nOFCsIIfWpP/jrfo0s"',
-    "mtime": "2026-06-30T09:01:48.556Z",
+    "mtime": "2026-07-01T02:23:48.448Z",
     "size": 11256,
     "path": "../public/assets/zhipuai-BPNAnxo-.svg"
   },
+  "/assets/_sessionId-BhNTuZ31.js": {
+    "type": "text/javascript; charset=utf-8",
+    "etag": '"d2-N3jVPi19Muv+Rgqnids/dlwNuhs"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 210,
+    "path": "../public/assets/_sessionId-BhNTuZ31.js"
+  },
   "/assets/qwen-CONDcHqt.png": {
     "type": "image/png",
     "etag": '"572c3-cdJAPaHdOvFCGzuaQjagdgOu6XE"',
-    "mtime": "2026-06-30T09:01:48.556Z",
+    "mtime": "2026-07-01T02:23:48.448Z",
     "size": 357059,
     "path": "../public/assets/qwen-CONDcHqt.png"
   },
-  "/assets/ProxyViewerContainer-D650kQlv.js": {
+  "/assets/ProxyViewerContainer-BxRaXhKR.js": {
     "type": "text/javascript; charset=utf-8",
-    "etag": '"af8bd-kLHfI4KjZ80Xpnq5/VoFdzyBH10"',
-    "mtime": "2026-06-30T09:01:48.557Z",
-    "size": 719037,
-    "path": "../public/assets/ProxyViewerContainer-D650kQlv.js"
+    "etag": '"b489b-R0CduB+BYfBdDePPpvc2Nkavq3A"',
+    "mtime": "2026-07-01T02:23:48.450Z",
+    "size": 739483,
+    "path": "../public/assets/ProxyViewerContainer-BxRaXhKR.js"
   }
 };
 function readAsset(id) {

package/README.md

@@ -154,7 +154,7 @@ http://localhost:25947/api/mcp
 Coding agents can use the MCP surface in three ways:
 
 - Tools: perform actions such as listing logs, reading bounded log details, testing providers,
-  creating evaluation runs, and exporting evidence packs.
+  creating evaluation runs/groups, attaching sessions, and exporting evidence packs.
 - Resources: read stable evidence URIs without invoking a tool call.
 - Prompts: start common analysis/reporting tasks from reusable templates.
 
@@ -176,6 +176,24 @@ The evaluation-oriented run flow is:
 <dataDir>/evidence/<runId>/evidence.html
 ```
 
+For benchmark or CI systems that spawn multiple coding-agent processes, use a group as the
+outer evaluation artifact:
+
+1. Call `inspector_create_group` once for the evaluation batch.
+2. Start each OpenCode/Codex process through the proxy with its own model/provider.
+3. As each process discovers or reports its session id, call `inspector_add_group_session` with
+   `groupId`, `sessionId`, and optional `runId`, `provider`, `model`, `agent`, `label`, and
+   metadata.
+4. Call `inspector_update_group` to mark the whole batch `running`, `failed`, `completed`, or
+   `cancelled`.
+5. Call `inspector_export_group_evidence` to write an aggregate provider/model matrix under:
+
+```text
+<dataDir>/evidence/groups/<groupId>/evidence.json
+<dataDir>/evidence/groups/<groupId>/evidence.md
+<dataDir>/evidence/groups/<groupId>/evidence.html
+```
+
 Useful MCP resource URIs:
 
 | Resource | Purpose |
@@ -187,12 +205,17 @@ Useful MCP resource URIs:
 | `inspector://failures/recent` | Read recent failed/cancelled runs and non-success evidence classifications. |
 | `inspector://runs/{runId}` | Read one declared run and its evidence paths. |
 | `inspector://evidence/{runId}` | Read the exported Markdown evidence pack for a run. |
+| `inspector://groups` | List declared evaluation groups. |
+| `inspector://groups/{groupId}` | Read one group, including attached session/run members and evidence paths. |
+| `inspector://groups/{groupId}/sessions` | Read only the attached group member list. |
+| `inspector://groups/{groupId}/evidence` | Read the exported Markdown evidence pack for a group. |
 
 Built-in prompt templates help turn captured evidence into user-facing output:
 
 - `inspector_analyze_session_failure`: explain why a session failed and cite log evidence.
 - `inspector_triage_recent_failures`: prioritize recent failed runs and suggest next actions.
 - `inspector_generate_jenkins_report`: produce a Jenkins-ready Markdown report for a run.
+- `inspector_generate_group_report`: produce a provider/model evaluation report for a group.
 - `inspector_extract_repro_steps`: extract concise reproduction steps from a session.
 
 ## Jenkins And GitCode

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tonyclaw/agent-inspector",
-  "version": "2.0.29",
+  "version": "2.0.31",
   "type": "module",
   "description": "Agent observability and knowledge capture layer for AI coding tools.",
   "license": "MIT",

package/src/components/ProxyViewer.tsx

@@ -15,6 +15,7 @@ import { crabVariants } from "./ui/crab-variants";
 import { McpLogo } from "./ui/mcp-logo";
 import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from "./ui/select";
 import { Tooltip, TooltipContent, TooltipProvider, TooltipTrigger } from "./ui/tooltip";
+import { GroupsDialog } from "./groups/GroupsDialog";
 import { SettingsDialog } from "./providers/SettingsDialog";
 import { computeCacheTrends } from "./proxy-viewer/cacheTrend";
 import { LazyCompareDrawer } from "./proxy-viewer/lazy";
@@ -38,6 +39,10 @@ function truncateSessionId(id: string): string {
   return `${id.slice(0, 12)}...${id.slice(-12)}`;
 }
 
+export function shouldShowRawExport(captureMode: CaptureMode): boolean {
+  return captureMode === "full";
+}
+
 function computeTokenSummary(logs: CapturedLog[]): { totalIn: number; totalOut: number } {
   let totalIn = 0;
   let totalOut = 0;
@@ -709,6 +714,7 @@ export function ProxyViewer({
   );
   const cacheTrends = useMemo(() => computeCacheTrends(groups), [groups]);
   const comparisonPredecessors = useMemo(() => buildValidPredecessors(groups), [groups]);
+  const showRawExport = shouldShowRawExport(captureMode);
   const handleCompareWithPrevious = useCallback(
     (log: CapturedLog) => {
       const predecessor = comparisonPredecessors.get(log.id);
@@ -788,7 +794,8 @@ export function ProxyViewer({
             </span>
             <McpReadyBadge />
           </h1>
-          <div className="justify-self-end">
+          <div className="justify-self-end flex items-center gap-1">
+            <GroupsDialog />
             <SettingsDialog />
           </div>
         </div>
@@ -874,7 +881,7 @@ export function ProxyViewer({
                 : ""}
             </span>
           )}
-          {logs.length > 0 && (
+          {logs.length > 0 && showRawExport && (
             <button
               type="button"
               onClick={() => {