AI Coding Session Log Viewer & Manager

What you can do with a session log

• Organize sessions: open logs from Claude Code, Cursor, Copilot CLI and Codex together. The file list groups them by agent and then by project, so a mixed collection stays easy to scan.
• Read the conversation: Markdown, syntax-highlighted code and copy buttons, with prompts on one side and the agent's replies on the other.
• Inspect tool calls: shell commands and their output, file edits as diffs, reads, searches, sub-agents and todo lists, each in a collapsible card.
• Search across sessions: every opened log is searched at once, with match counts per file and jump-to-match navigation.
• Follow the session: a clickable outline of prompts, plus model, branch, working directory, duration, tokens and cost when the log records them.
• Follow a live session: turn on Live and the open log refreshes as the agent writes new messages, so you can watch the current working session without reloading by hand. Tool cards render lazily to keep long logs fast.
• Control the view: show or hide tool calls, injected context and the agent's thinking, and expand or collapse every tool card at once.
• Export a session: save it as a self-contained, print-friendly HTML file or as Markdown for sharing or archiving.

How to open and search session logs

1. Find the .jsonl file written by your agent. The common locations are listed above.
2. Drag it onto the page, or click Add files (or Add folder) in the Files panel.
3. Click the file in the list to open it. Press / or Ctrl+F to search, then Enter and Shift+Enter to move between matches.

Chrome and Edge can remember opened files across page reloads and re-read them from disk. Switch on Live next to Reload and the open file is re-read whenever it changes on disk, which keeps a running session up to date without manual reloads. Firefox and Safari work through the regular file picker, where the list does not survive a reload.

Privacy

Session logs are read and rendered locally in your browser. The viewer never uploads file contents, and there is no server-side processing. File handles and preferences are stored only in your browser's IndexedDB and localStorage.

The page loads a lightweight Umami page-view counter. It records page visits only and never receives the contents of your logs.

Feedback is collected with Tally, a third-party form service. The form only loads after you click a feedback link. Tally receives the page URL and the details you enter, including the name, email, message and any file you choose to attach, and may set its own cookies. This is separate from the logs you view: opening a log never uploads it. If you attach a log to a bug report, only that file is sent to Tally, so remove secrets and personal data first.

Frequently asked questions

Are my session logs uploaded?

No. The viewer opens files with the browser's file APIs and renders them on your device; it never uploads them. A log is only sent if you choose to attach one to a feedback report.

Which coding agents are supported?

Claude Code, Cursor, GitHub Copilot CLI and Codex. The format is detected automatically from the file content.

Can I view logs from different agents together?

Yes. Open logs from any supported agent in the same list. They are grouped by agent and project, and cross-file search covers all of them, which makes it easy to compare how different agents handled the same task.

Can I request support for another agent?

Yes. Use the feedback form to request a log format or report an issue. Attach a sample log only if it helps, and remove secrets and personal data first.

Where do the log files live?

Claude Code writes to ~/.claude/projects/, Cursor to ~/.cursor/projects/, Copilot CLI to ~/.copilot/session-state/, and Codex to ~/.codex/sessions/.

Can I open a whole folder?

Yes. In Chrome and Edge, Add folder walks up to four levels deep and accepts up to 1,000 files, which suits folder layouts that nest sessions by project or date.

Can I follow a session that is still running?

Yes. Open the log in Chrome or Edge and turn on Live. LogTrove watches the file and re-reads it when the agent appends new messages, so the view stays current as you work. The status shows when live refresh is paused during in-page search, inline rename or while the tab is in the background, then catches up when you return.

Does it work offline?

The viewer is a static page. Once it has loaded, opening and searching local logs does not need a network connection.

Can I export a session?

Yes. Open a log and choose HTML, Markdown or Copy Markdown from the Export menu in the bottom-right corner. The menu also controls which Tools / Context / Thinking blocks are included and the theme of the exported HTML, which is self-contained and works offline.

Feedback & agent requests

Tell us which agent log format you need next, or report a problem with a log. The form is in English, hosted by Tally, and opens in the bottom-right corner without changing the page.

Send feedback