refactor(runtime): enforce source file size boundaries

src/main/agent-runtime/artifact-baseline.ts

@@ -0,0 +1,97 @@
+import { relative, resolve } from "node:path";
+import { readdirSync, statSync } from "node:fs";
+import { isGitRepo } from "./artifact-git";
+import { logArtifactFsError, type Logger } from "./artifact-logging";
+import type { WorkspaceBaseline } from "./artifact-types";
+
+const MAX_BASELINE_ENTRIES = 20_000;
+
+export const BASELINE_SKIP_DIR_NAMES: ReadonlySet<string> = new Set([
+  ".git",
+  "node_modules",
+  ".next",
+  "dist",
+  "build",
+  "out",
+  ".cache",
+  ".pnpm-store",
+]);
+
+/**
+ * Capture a run-start mtime baseline only for non-git workspaces.
+ * Git workspaces use `git status` as the authoritative diff source.
+ */
+export function captureBaselineIfRequired(
+  workspacePath: string | null,
+  logger?: Logger,
+): WorkspaceBaseline | null {
+  if (!workspacePath) return null;
+  if (isGitRepo(workspacePath)) return null;
+  try {
+    if (!statSync(workspacePath).isDirectory()) return null;
+  } catch (error) {
+    logArtifactFsError(
+      logger,
+      error,
+      "artifact.fs.workspace_stat_failed",
+      "Failed to stat workspace before artifact baseline capture.",
+      { workspacePath },
+    );
+    return null;
+  }
+
+  const mtimeByRelPath = new Map<string, number>();
+  const stack: string[] = [workspacePath];
+  let truncated = false;
+  while (stack.length > 0) {
+    if (mtimeByRelPath.size >= MAX_BASELINE_ENTRIES) {
+      truncated = true;
+      break;
+    }
+    const dir = stack.pop();
+    if (!dir) break;
+    let entries: { name: string; isDirectory: () => boolean; isSymbolicLink: () => boolean }[];
+    try {
+      entries = readdirSync(dir, { withFileTypes: true });
+    } catch (error) {
+      logArtifactFsError(
+        logger,
+        error,
+        "artifact.fs.baseline_readdir_failed",
+        "Failed to read workspace directory during artifact baseline capture.",
+        { workspacePath, dir },
+      );
+      continue;
+    }
+    for (const entry of entries) {
+      const abs = resolve(dir, entry.name);
+      if (entry.isSymbolicLink()) continue;
+      if (entry.isDirectory()) {
+        if (BASELINE_SKIP_DIR_NAMES.has(entry.name)) continue;
+        stack.push(abs);
+        continue;
+      }
+      try {
+        const stat = statSync(abs);
+        if (!stat.isFile()) continue;
+        const rel = relative(workspacePath, abs);
+        if (rel.length === 0) continue;
+        mtimeByRelPath.set(rel, stat.mtimeMs);
+      } catch (error) {
+        logArtifactFsError(
+          logger,
+          error,
+          "artifact.fs.baseline_stat_failed",
+          "Failed to stat workspace file during artifact baseline capture.",
+          { workspacePath, absPath: abs },
+        );
+        continue;
+      }
+      if (mtimeByRelPath.size >= MAX_BASELINE_ENTRIES) {
+        truncated = true;
+        break;
+      }
+    }
+  }
+  return { mtimeByRelPath, capturedAt: Date.now(), truncated };
+}