Show HN: Claude-replay – A video-like player for Claude Code sessions(github.com)
github.com
Show HN: Claude-replay – A video-like player for Claude Code sessions
https://github.com/es617/claude-replay
39 comments
I'm curious what people here use for sharing agent coding sessions with colleagues? I'm sure we can learn from each other, but there's no perfect tool.
The ones I've used which can convert coding session histories into readable HTML are:
https://github.com/simonw/claude-code-transcripts (only works with Claude Code)
https://github.com/Dicklesworthstone/coding_agent_session_se... (supports many coding agents but the tui feels kind of clunky and it only exports one session at a time)
The ones I've used which can convert coding session histories into readable HTML are:
https://github.com/simonw/claude-code-transcripts (only works with Claude Code)
https://github.com/Dicklesworthstone/coding_agent_session_se... (supports many coding agents but the tui feels kind of clunky and it only exports one session at a time)
I would have made a "claude2asciinema", for geek points. Also an opencode version.
Ha, I like the claude2asciinema name!
I did think about asciinema-style output, but I wanted the player to expose the structured steps rather than just replay the terminal output. Supporting other agent CLIs could definitely be interesting.
I just added a Claude alias that calls Claude with flags wrapped in asciinema. Only annoying thing is that people have wanted video or gifs and the conversion has been annoying a few times. Will fix it later.
This is amazing for getting new team members onboarded with Claude Code processes and tools I use. Thanks for working on it. will give it a try tonight.
Neat. Would be nice if I could easily drop the replayer into a slack thread. Maybe a video file would be better for that than an HTML file?
Good point. The reason I went with HTML is that you can expand tool calls, inspect prompts, jump around the timeline, etc., which would be hard to preserve in a video.
But you’re right that sharing in places like Slack could be easier with a video preview. Something to think about.
But you’re right that sharing in places like Slack could be easier with a video preview. Something to think about.
Question: how can you find the exact session you are looking for, among hundreds of them? I had a look at my ~/.claude/projects/*/ and I couldn't even find my last session.
I had exactly this problem and didn’t see anything good out there (Claude —resume
only searches session names and auto-created titles) so I got a tool built that uses a Rust/Tantivy full text search index. It’s part of the aichat command suite, called “aichat search”:
https://pchalasani.github.io/claude-code-tools/tools/aichat/...
It brings up a nice TUI for filtering and further actions. There’s also a —json flag so agents can use it as a CLI search tool to find context about any past work. There’s a plugin that provides a corresponding session-searcher agent that knows to use this tool to search sessions.
I have hundreds/thousands of past sessions and this has been a life saver; I can just ask the main agent, “use the session searcher agent to get the details of how we built the tmux-cli tool so we can add some features”.
https://pchalasani.github.io/claude-code-tools/tools/aichat/...
It brings up a nice TUI for filtering and further actions. There’s also a —json flag so agents can use it as a CLI search tool to find context about any past work. There’s a plugin that provides a corresponding session-searcher agent that knows to use this tool to search sessions.
I have hundreds/thousands of past sessions and this has been a life saver; I can just ask the main agent, “use the session searcher agent to get the details of how we built the tmux-cli tool so we can add some features”.
Ha, good question. Short answer: I often let Claude Code find it.
Sessions are grouped by the folder where you ran Claude Code (e.g. ~/.claude/projects/Users-<user>-<path>), so if you don’t run everything from the same directory, it’s usually easy to narrow down.
They’re also plain JSONL files, so grep works well if you remember part of a prompt.
That said, it might be nice for claude-replay to add a helper command to list or search recent sessions.
Sessions are grouped by the folder where you ran Claude Code (e.g. ~/.claude/projects/Users-<user>-<path>), so if you don’t run everything from the same directory, it’s usually easy to narrow down.
They’re also plain JSONL files, so grep works well if you remember part of a prompt.
That said, it might be nice for claude-replay to add a helper command to list or search recent sessions.
In your last session, use "/status" to show your session_id, then find your session file in "~/.claude/projects/[your_project]/[session_id].jsonl"
How are people sharing reproducible AI coding sessions today? Screenshots and recordings never capture the full interaction.
Does it work when using Claude inside Cursor? Or is there some something like claude-replay but for Cursor sessions? (I'm out of home, can't try now)
good question. Currently, it works with standalone Claude code or Claude code via VSC extension (both terminal and chat).
Cursor stores JSONL logs, too, but no tool calls or timestamps. I'll add support for it soon; it will be a bit less informative, and playback pacing will be heuristic, but it's a good addition.
Hi, I pushed a new version with Cursor support. Cursor logs don’t include explicit thinking blocks or timestamps, so the player estimates both heuristically. Intermediate assistant messages are treated as “thinking” and timing is approximated for playback. Let me know if you try it!
Do you think this can be used as an observability tool? does claude code logs the LLM reasoning?
This is great. Excellent for knowledge sharing sessions and internal trainings. Thank you for putting this together so my clanker doesn't have to!
I can’t say I understand why one would want this lol. Watching cc session replays doesn’t seem particularly useful. But the execution seems well done, so nice job!
I'm discovering new possibilities all the time with how Claude can work on a new type of task in our codebase and business more broadly. While a lot of this can be brought to the team by saying "encapsulate what you just did into a skill," sometimes it's as much about knowing what kinds of prompts to use to guide it as well.
Showing a colleague that flow, and the sequence of not just prompts but the types of Claude outputs to expect, all leading to Claude doing something that would have taken us a half day of work? As a linear video, rather than just a dump or screenshot of a final page? That could help to diffuse best practices rapidly.
OP - you might want to look at the kind of data model Loom used for this problem for videos in general, in terms of workspaces and permissions. Could make a startup out of this!
(Also as a smaller note - you might want to skip over long runs and generations by default, rather than forcing someone into 5x mode! A user of this would want to see messages, to and from Claude, at a standardized rate - not necessarily a sped up version of clock time.)
Showing a colleague that flow, and the sequence of not just prompts but the types of Claude outputs to expect, all leading to Claude doing something that would have taken us a half day of work? As a linear video, rather than just a dump or screenshot of a final page? That could help to diffuse best practices rapidly.
OP - you might want to look at the kind of data model Loom used for this problem for videos in general, in terms of workspaces and permissions. Could make a startup out of this!
(Also as a smaller note - you might want to skip over long runs and generations by default, rather than forcing someone into 5x mode! A user of this would want to see messages, to and from Claude, at a standardized rate - not necessarily a sped up version of clock time.)
That’s a really interesting way to frame it — showing the flow of prompts and responses rather than just the final result.
I’ve mostly been using it for demos and sharing sessions with teammates, but the training / best-practices angle is a great point.
On navigation: you can already step through turns with the arrow keys or jump around the timeline, so you don’t have to sit through long generations. But I agree that smarter defaults (skipping or collapsing long runs) could make it smoother.
And the Loom comparison is interesting — I hadn’t thought about the workspace/permission side yet since this started as a small CLI tool for sharing sessions, but that’s a good direction to think about.
I’ve mostly been using it for demos and sharing sessions with teammates, but the training / best-practices angle is a great point.
On navigation: you can already step through turns with the arrow keys or jump around the timeline, so you don’t have to sit through long generations. But I agree that smarter defaults (skipping or collapsing long runs) could make it smoother.
And the Loom comparison is interesting — I hadn’t thought about the workspace/permission side yet since this started as a small CLI tool for sharing sessions, but that’s a good direction to think about.
> Showing a colleague that flow, and the sequence of not just prompts but the types of Claude outputs to expect, all leading to Claude doing something that would have taken us a half day of work? As a linear video, rather than just a dump or screenshot of a final page? That could help to diffuse best practices rapidly.
Would this not be visible in a text dump without taking half a day to watch? What's/who's the benefit/benificiary of the realtime experience here?
Granted, I have friends who don't read but prefer visual stimulation. I don't think the overlap with people comfortable with code is very large at all.
Would this not be visible in a text dump without taking half a day to watch? What's/who's the benefit/benificiary of the realtime experience here?
Granted, I have friends who don't read but prefer visual stimulation. I don't think the overlap with people comfortable with code is very large at all.
They can go hand in hand! But if you give a dump of a session to someone, with literal reams of command inputs and outputs etc. interleaved in the session... they'll most likely read the beginning and the end. And possibly absorb the importance of the discovery, but not the types of "prodding" I was doing to the agent to make that possible.
Slowing it down to show the back-and-forth, and to let the viewer absorb and internalize the techniques behind each "prod," is vital!
Slowing it down to show the back-and-forth, and to let the viewer absorb and internalize the techniques behind each "prod," is vital!
I thought the same about watching people play video games but that's clearly a thing! This might be useful for educating people on how to use these new tools, perhaps those not in engineering but product, UX, less familiar with CLIs.
Thank you, and fair question :) I’ve been experimenting with using Claude Code with hardware, where the interesting part is the tool usage and workflow, not just the final output. Screenshots and recordings made it hard to show, so the replay lets you step through the session and inspect what actually happened.
I think the main use case is training. I feel more and more confident with my prompts ( and what tasks I can safely pass to what models ), but it is sometimes hard to explain to anyone else what made me go a particular route. This may help, because a person can follow your intuition.
cool project!
related: I made https://github.com/clkao/agentlore can aggregate logs across team, and sharable with permalink. the logs are processed by agentsview, which supports multiple coding agents.
related: I made https://github.com/clkao/agentlore can aggregate logs across team, and sharable with permalink. the logs are processed by agentsview, which supports multiple coding agents.
Very nice, thank you. This will come in handy for a scientific agent I am working on.
Thanks! That sounds like a great use case. If you try it with your agent, let me know how it works out.
really need tools like this, show how powerful my skill is to my colleagues
Nice to see a AI coding tool that not (complete) vibe coded! Well done.
Looking forward to try this with my students. Thanks!
Excellent share! Danke.
This is really nice.
aiaboratory(1)
Claude Code already stores full session transcripts locally as JSONL files. Those logs contain everything: prompts, tool calls, thinking blocks, and timestamps.
I built a small CLI tool that converts those logs into an interactive HTML replay.
You can step through the session, jump through the timeline, expand tool calls, and inspect the full conversation.
The output is a single self-contained HTML file — no dependencies. You can email it, host it anywhere, embed it in a blog post, and it works on mobile.
Repo: https://github.com/es617/claude-replay
Example replay: https://es617.github.io/assets/demos/peripheral-uart-demo.ht...