Spotlight by Backplanes - Session reports for Claude Code & Codex to improve your code
by•
Keep up with your agents. Spotlight reads your Claude Code and Codex sessions and shows you what your agents actually did, and how to get recursively better every session: what to fix now, what to ship better next time, what's worth sharing. One harness or seven, solo or across your team. Free.
Replies
Lancepilot
@odeth_negapatan1 honest answer: the most eye-opening pattern wasn't actually a risk. When we stitched our own sessions together, the scary stuff was there (an API key in a tracked file started this product), but the bigger surprise was how many good moves never traveled. One of us had solved a problem cleanly, in one session, on one machine, and the rest of us kept burning time on the same thing for weeks. Nobody was hiding anything; there was just no surface for the unlock to move across. That discovery was a genuine unlock for us, and is a core part of why Spotlight focuses so heavily on highlighting effective patterns worth sharing..
What I've found fascinating is almost every first report we've watched land in person contains one thing the person would have sworn didn't happen. That moment is kind of the product. Curious what yours catches, if you give it a run. :)
Fella
Interesting history for how you stumbled on the product. Excited to try.
Spotlight by Backplanes
@richie_cartwright thanks, Richie! Equally excited to hear what you find and what we can improve for you!
Congratulations, @gogogadgetneil and team. This is much needed - way more than many realize. My experience was similar to what Neil shared on the LinkedIn post about unseen changes made or even files read outside a project being worked on.
Spotlight by Backplanes
@gogogadgetneil @srinath_murthy we couldn't agree more! We hear this ad nauseam from everyone we talk to, and are so glad to help people meet this need. Would love to learn from you what connects most about the reports, and what work you'd love to see tackled next!
Octane AI
Congrats on the launch! Excited to see how this evolves.
Spotlight by Backplanes
@benparr thank you! Super curious what you learn from Spotlight and how we can make it better for you!
As an engineer who gets flak for not selecting "auto accept edits" and wanting to observe and approve most (if not all) actions Claude does, I love this idea. This kind of insight is sorely needed.
Spotlight by Backplanes
@dan_dias You're the engineer the rest of us pretend to be 😄 And your instinct is right: paying that much attention is the only honest option today, and you still miss things (I bet you're not watching all the subagent tool calls, are you?). There has to be a better way -- the speed without the babysitting, where you see and manage what actually matters instead of approving every single action. That's the bet we're making. Try it and tell us if it lets you loosen the grip a little because you can now catch what matters without needing to inspect it all. And then we'd love to hear what you want from us next!
The session-report angle for Claude Code/Codex is pretty useful. I might have missed it, but do you separate “what changed” from “why the agent made that choice”? That distinction matters a lot when reviewing messy agent runs.
Spotlight by Backplanes
@xiaosong001 Thanks for the question. This is a good one! We capture reasoning and subagent behavior in our analysis of the session. We know the context of response/reasoning/subagent/tool is really important to understand why something became messy or went off the rails. That's why we make sure that the entire session is processed (or at least processed as very large chunks for you hardcore people with 300h+ sessions). Thanks for your comment, and the insight!
Spotlight by Backplanes
What a launch day. THANK YOU, Product Hunt. 🔦
#3 Product of the Day, and honestly the part we'll remember is this comment section. You showed up with war stories, hard questions, and lines better than our own copy ("invisible until it isn't" -- still thinking about that one).
Huge thanks to @fmerian for hunting us and being everywhere in this thread all day, to @chrismessina for the support and one very relatable .env confession, and to the @Spotlight by Backplanes team -- @gogogadgetneil, @nick_vlku, Andrew, and Christina -- what you pulled off this week was incredible.
And we heard you on what to build next:
More harnesses: Cursor, Cowork, Copilot
Windows support
Deeper reads on quality of work, and the "why" behind agent choices
Catching the things agents SAY are done but aren't
Automating the insights, so your agents get recursively better without you lifting a finger
Keep telling us what you wish you could see -- we want to build this all for you, and we read every comment you make. 🔦
The SSH-key story is exactly the gap. Session reports show the blast radius after the fact. Are you thinking about pre-action gates too, like: this agent can read these paths, write only there, and ask before secrets or env files?
Spotlight by Backplanes
@blah_mad 1000% that's exactly the roadmap! Help you set policy so that these types of things can be prevented or steered in the right direction, and to create an automated feedback loop between Spotlight and your coding harness to self-improve in the manner you want.
@antifreeze yes, that loop is the sharp part. The tricky bit is separating learn from this session from change policy automatically. I would want a small approval gate there, otherwise the safety layer becomes another agent with permissions.
The credential and attribution side is well covered here, so a different angle. The loop only really closes if a report changes the next session, not just the engineer who reads it. Do you feed findings back into the harness as prospective guardrails, a CLAUDE.md rule or a Codex instruction, so the agent stops re-making the same mistake? Or is improvement still mediated entirely by a human remembering the report afterwards? The retrospective is useful, but the prospective write-back is where this gets compounding.