feat(init): report repository scan omissions

[betegon]

The init workflow currently receives only files that happened to be discovered. It cannot tell whether a scan was complete, which directories were intentionally not traversed, or why a requested file came back as null.

Summary

• Add backward-compatible directory scan metadata with an explicit truncated flag
• Report visible directories omitted because they are hidden, excluded, beyond max depth, or seen during a non-recursive scan
• Report a per-file read result: read, truncated, skipped, or error, including bounded byte counts and stable failure reasons
• Carry initial scan metadata into workflow state and tag truncation plus omitted-directory count on CLI telemetry
• Preserve the existing entries and files fields for older servers
• Keep agent-selected binary paths readable; only automatic pre-reads are filtered by the preceding PR

Stack

• Based on fix(init): skip binary automatic pre-reads #1402, which filters binary content only from automatic common-file pre-reads.
• The server-side consumer is getsentry/cli-init-apifix(upgrade): handle EPERM in isProcessRunning for cross-user locks #211. It puts scan gaps and read outcomes into the untrusted repository context used by discovery, detection, planning, and installation agents.
• The readdir propagation fix is isolated in the follow-up fix(init): surface directory read errors #1405.

Test plan

• 83 targeted init scanner, filesystem tool, and wizard-runner tests
• Biome check on changed files
• pnpm typecheck

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
cli	Ready	Preview	Aug 11, 2026 1:11pm