chore: v2.8.13 — Artifacts reinstall/edit/installs, agent steps in chat, pick a local file to attach, Windows paths as links

- Chat: an Agent step shows its task, status and tool calls and opens its conversation in place; Windows paths (E:\…, C:/…) in replies open as files
- Prompt: attach a file on this computer by path (native picker), clearer too-large messages
- Artifacts: edit title/description and a skill's files (new snapshot), Reinstall, install history, Hub-only rows, skills installed into ~/.claude/skills and ~/.agents/skills
- Files: batch upload of small files in one extract
- AI Tools: npm install falls back to --prefix ~/.local when the global prefix is not writable

server/routes/sessions.js

@@ -846,7 +846,13 @@ router.get("/:id/transcripts", async (req, res) => {
     const omp = require("../lib/omp-usage");
     const list = [];
     if (omp.journalPath(req.params.id)) {
-      list.push({ id: "main", name: "Main Agent", type: "main", has_transcript: true, db_agent_id: null });
+      list.push({
+        id: "main",
+        name: "Main Agent",
+        type: "main",
+        has_transcript: true,
+        db_agent_id: null,
+      });
     }
     for (const j of omp.subagentJournals(req.params.id)) {
       list.push({
@@ -973,6 +979,8 @@ router.get("/:id/transcripts", async (req, res) => {
           name: transcriptName,
           type: isCompact ? "compaction" : "subagent",
           subagent_type: transcriptSubagentType,
+          // The Agent call that launched it: the chat opens it from that call.
+          tool_use_id: typeof meta?.toolUseId === "string" ? meta.toolUseId : null,
           has_transcript: true,
           db_agent_id: null, // matched later after all transcripts are collected
           _timestamp: transcriptTimestamp,
@@ -1260,18 +1268,32 @@ router.get("/:id/transcript", async (req, res) => {
               // A line carries one tool result; an edit's +/- counts ride on it.
               const diff = patchCounts(entry.toolUseResult);
               const answers = askAnswers(entry.toolUseResult);
+              const agent = agentInfo(entry.toolUseResult);
+              // A subagent's report is a reply, read as prose: its text blocks
+              // whole, not a 10 KB slice of their JSON.
+              const report =
+                agent && Array.isArray(block.content)
+                  ? block.content
+                      .filter((c) => c?.type === "text" && typeof c.text === "string")
+                      .map((c) => c.text)
+                      .join("\n\n")
+                  : null;
               content.push({
                 type: "tool_result",
                 id: block.tool_use_id || null,
-                output: truncate(
-                  typeof block.content === "string"
-                    ? block.content
-                    : JSON.stringify(block.content || ""),
-                  10240
-                ),
+                output:
+                  report !== null
+                    ? truncate(report, TEXT_MAX)
+                    : truncate(
+                        typeof block.content === "string"
+                          ? block.content
+                          : JSON.stringify(block.content || ""),
+                        10240
+                      ),
                 is_error: !!block.is_error,
                 ...(diff ? { diff } : {}),
                 ...(answers ? { answers } : {}),
+                ...(agent ? { agent } : {}),
               });
             }
           }
@@ -1574,7 +1596,10 @@ function ompMessage(entry, num, isSubagent = false) {
             type: "tool_use",
             name: c.name || "unknown",
             id: c.id || null,
-            input: truncateObj(c.arguments && typeof c.arguments === "object" ? c.arguments : {}, 10240),
+            input: truncateObj(
+              c.arguments && typeof c.arguments === "object" ? c.arguments : {},
+              10240
+            ),
           });
         }
       }
@@ -1670,6 +1695,24 @@ function askAnswers(r) {
   return Object.keys(out).length ? out : null;
 }
 
+/** An Agent/Task result's facts (`toolUseResult`): the subagent's id (its
+ *  transcript is `agent-<id>.jsonl`), the model it ran on and, once it
+ *  finished in the foreground, its time, tokens and tool calls. Null for any
+ *  other result. */
+function agentInfo(r) {
+  if (!r || typeof r !== "object" || typeof r.agentId !== "string") return null;
+  if (!/^[\w-]{1,200}$/.test(r.agentId)) return null;
+  const num = (v) => (Number.isFinite(v) ? v : null);
+  return {
+    id: r.agentId,
+    model: typeof r.resolvedModel === "string" ? r.resolvedModel : null,
+    status: typeof r.status === "string" ? r.status : null,
+    duration_ms: num(r.totalDurationMs),
+    tokens: num(r.totalTokens),
+    tool_uses: num(r.totalToolUseCount),
+  };
+}
+
 // A conversation turn — a prompt, an answer — is shown whole: it is bounded by
 // the model's output and what the user typed. Only tool payloads and thinking
 // (the parts that can run to megabytes) keep the 10 KB cap. This one guards
@@ -1694,3 +1737,4 @@ module.exports.sessionStats = sessionStats;
 module.exports.classifyTranscriptSender = classifyTranscriptSender;
 module.exports.patchCounts = patchCounts;
 module.exports.askAnswers = askAnswers;
+module.exports.agentInfo = agentInfo;