![Cover image for Share, Embed, and Curate Agent Sessions on DEV [Beta]](https://media2.dev.to/dynamic/image/width=1000,height=420,fit=cover,gravity=auto,format=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fwk6kbgjbme42vp7aotnx.png)
Authors can now upload Agent Session transcripts to their DEV account. Those transcripts can be curated, sliced up, shared publicly, and embedded i...
For further actions, you may consider blocking this person and/or reporting abuse
The slice-before-upload gap @pascal_cescato_692b7a8a20 identified is the right one. for embedding in technical writing, the useful unit is usually a specific exchange. The moment the agent made the wrong decision or the prompt that finally produced the right output. uploading a full session to get access to two minutes of it adds friction that'll stop most writers from using the feature at all.
Building something adjacent — a browser extension for capturing Claude conversations with semantic search and MCP access. The portable conversation artifact problem is the same one you're solving here, different layer.
will test with a Claude Code session.
Will see what we can do here :)
Based on your and Pascal's feedback I just shipped a major refactor to handle more of the curation/processing on the client side and remove the 10MB limit as a result. Curious to hear your thoughts!
That's a fast turnaround. will test with a Claude Code session next week and report back.
Removing the 10MB limit by moving curation client-side is the right architectural call. The constraint was in the wrong place. The useful unit for embedding in technical writing is a specific exchange, not a full session, so client-side selection before upload makes more sense than upload-then-slice anyway.
It's an awesome feature! I've discover it this afternoon, and tested… I'm found of!
Awesome - have you used it in a post yet? Would love to check it out if so
Unfortunately not yet. This feature arrived a bit too late for the article where I could have used it… and my sessions (GitHub Copilot CLI) exceed the 10MB limit. That said, it's definitely on my radar for a future article!
Do you think that limit will generally be too low?
The limit itself may not be the core issue — if I had known about it from the start, I would have managed my sessions differently: intermediate saves, one session per feature. The real pain was discovering it after the fact, with a session already too large to use.
Also, for an article, you wouldn't embed a full 10MB session anyway — you'd want a meaningful excerpt. So the real need might be less about raising the limit and more about having a way to extract or slice a specific part of a session. Which, coincidentally, might be my next little experimentation… 😄
Good point. We have a slice feature but it's post upload. Maybe there's a way to move it earlier in the process
Exactly — the slices feature is great, but it's a post-upload tool. If the session exceeds 10MB, you never get past the upload step, so slicing never becomes an option. Moving that capability earlier in the process would be the key!
Just shipped a major refactor to handle more of the curation/processing on the client side and remove the 10MB limit as a result. Curious to hear your thoughts!
I just tried, works fine! So I've included some exchanges with Github Copilot CLI here: dev.to/pascal_cescato_692b7a8a20/a.... Thanks @jonmarkgo!
Love it!
This is exciting! Agent session sharing could be huge for collaborative debugging. I run multi-agent systems where sub-agents handle different tasks and being able to share a specific agent's session with a teammate would save hours of 'let me explain what happened'. Will this support embedding in external docs?
External embedding is not supporting by default, but happy to look into that if you want to open a feature request Issue on GitHub github.com/forem/forem/issues
It's awesome. Thanks👌
this is great!
thinking of ideas to use the agent sessions :)
This is my guinea pig Pearl. She's not very good at coding yet.
