Spotlight by Backplanes

Key Features

• •Session Reports with Verdict: After each Claude Code or Codex session, Spotlight generates a report with a verdict like 'Needs review' or 'Business as usual'. It includes a time breakdown, findings, and a session story, giving you immediate insight into what your agent did. Benefit: quickly identify issues and validate successful operations without manual checking.
• •Auto-Capture and Local Redaction: Spotlight automatically captures sessions when they end and applies local redaction to strip PII and credentials before any data leaves your laptop. Benefit: maintain security and privacy with no need for OAuth integration, allowing you to stay in flow while ensuring sensitive information never leaves your machine.
• •Prioritized Dashboard: The dashboard organizes reports by status, with 'Needs review' reports highlighted so you can focus on critical sessions first. Examples include security flags like 'Agent opened /etc/passwd outside the task scope' and clear passes like 'Clean refresh, signature verified'. Benefit: save time by only reviewing sessions that require your attention.
• •Multi-Platform Support: Spotlight works with a variety of AI coding agents including Claude Code, Codex, Antigravity CLI, Google, OpenCode, Cursor, and more. It also supports macOS, Linux, and WSL 2. Benefit: use a single tool across different agent environments and operating systems without compatibility concerns.
• •Free for Individuals and Teams: Spotlight is free with no credit card required, no seat counting, and no trial clock. Individual developers and entire teams can use it immediately. Benefit: accessible for all sizes of development groups, from solo coders to large engineering organizations.
• •No OAuth Required: Unlike other monitoring tools, Spotlight does not require OAuth integration with Anthropic or OpenAI. The CLI only reads sessions after they end, using local redaction for security. Benefit: simple, secure setup that respects user privacy and avoids third-party authentication complexity.

Use Cases

• •Reviewing Agent Activity After a Meeting: Your AI agent ran for 47 minutes while you were in a meeting. Spotlight's session report shows a time breakdown, findings, and a verdict, allowing you to quickly assess what actions were taken, what to keep, and where to improve the prompt for next time. Outcome: faster post-meeting review and enhanced agent prompting.
• •Catching Security Mistakes: A password-reset session inadvertently opened /etc/passwd outside the task scope. The report is flagged 'Needs review', immediately bringing the security concern to your attention. Outcome: prevent insecure code from being merged and reinforce safety guardrails for future sessions.
• •Validating Code Refactoring: An auth-refresh-token session is marked as 'Clean refresh, signature verified'. The report confirms the agent behaved correctly and the refactor is safe. Outcome: confidence to merge without additional manual verification.
• •Verifying Database Migrations: A migrate-postgres-v14 session completes with 'Migration ran clean, no schema drift'. The report validates that the database changes were applied smoothly. Outcome: reduced risk during schema updates and faster release cycles.
• •Approving UX Tweaks: A ux-onboarding-tweak session updates three components and all tests pass. The 'Business as usual' report confirms the changes are safe and functional. Outcome: accelerate UI iteration with fewer manual checks.

Who is this for?

Developers using Claude Code or Codex for AI-assisted coding, including AI engineers, prompt engineers, and engineering managers who need visibility into agent sessions. Also suitable for teams adopting AI coding tools and wanting to review agent behaviour, improve prompts, and save time. Works on macOS, Linux, and WSL 2.