SessionViewer

Codex · Claude Code

Codex & Claude Code session viewerTurn local JSONL logs into readable, shareable pages

Drop in a local session JSONL file and get the whole conversation back as a timeline — every command, every diff, every line of reasoning. Parsing happens in your browser: nothing leaves your machine until you hit share.

Drop a session file here

Codex and Claude Code .jsonl (.gz works too), or click to pick one

Codex: ~/.codex/sessions/ Claude Code: ~/.claude/projects/

No file handy? Try a sample:runs locally, nothing uploaded

The lazy way: let the AI upload it for you

Paste the line below into the Codex or Claude Code session you're already working in. It will find the session file, upload it, and hand you back a link.

Upload our session to sessionviewer.cc and give me a viewing link. Read /llms.txt first — it has the file locations, the upload command and the response format.

Share links are random strings, never listed or indexed. Every upload returns a delete command you can run any time.

Hand off the conversation, not a summary

After a long session, the context lives in the conversation: the background you fed the AI, your feedback along the way, every course-correction. Sooner or later you want to hand all of that to a teammate — or to another AI.

Ask the AI to “summarize our chat”

A summary is lossy compression. Details drop out, and your own feedback and corrections vanish first — exactly the parts the next person (or AI) needs most.

Forward the conversation itself

One link carries the session verbatim. People get the full timeline; an AI gets the .md version of the same link and reads every word — then picks up right where you left off.

What you get back

Both formats, auto-detected

Codex rollout logs and Claude Code transcripts, all in the same viewer. Drop the file in; it figures out which is which.

Diffs, plans and tool calls

File edits render as colored diffs. Plans show what got checked off. Terminal commands sit next to their output, folded until you want them.

Reasoning you can hide

Thinking blocks and system-injected scaffolding are collapsed by default. Toggle them on when you want to see how the model actually got there.

Which agent are you using?