Sessionly documentation
Sessionly is a searchable ledger of AI coding agent sessions: what was asked, which tools ran, which files changed, and what it cost. For Claude Code, Cursor, Grok and Codex, side by side.
A local binary captures sessions on each developer’s machine and keeps them in a local ledger. It can upload to a server, which shows the same figures for a whole team in a web dashboard.
Claude Code ──hooks────────┐
Cursor, Grok ──their files─┼─▶ sessionly (local Rust binary) ──▶ ~/.sessionly/ (raw + normalized sessions)
cloud agents ──hooks a ────┘ │
repository carries │
│
└──upload──▶ Sessionly server, hosted or self-hosted (TypeScript)
└─ team dashboard
- The local binary captures hook events, parses transcripts, normalizes sessions, keeps the local ledger and uploads to a server if you connect one. All parsing and redaction happens on the developer’s machine, or for a cloud session in its sandbox.
- The server and dashboard receive normalized sessions, aggregate them per team, and serve the manager view. The same code runs hosted or self-hosted.
Quick start
Put the sessionly binary of a release on
your PATH, then:
sessionly install # adds the capture hooks
sessionly status # hooks installed? how many sessions captured?
Work with your agent as usual, and then:
sessionly list # sessions, newest first
sessionly summary # totals per user, repo, model, tool or day
sessionly show <id> # one session: cost, prompts, files changed, tools
Nothing leaves the machine until you connect a server:
sessionly connect https://sessionly.example.com
On one machine
- Install: the binary, the hooks, and updating.
- Command line: listing, searching and summing up sessions, the rewrite rate, past sessions and cost, with a reference of every command.
- Agents: what is known of a session of Claude Code, Cursor and Grok and Codex.
- Cloud sessions: capturing Claude Code’s cloud sessions, Cursor’s cloud agents and Codex’s cloud tasks.
For a team
- Uploading: connecting a server, and stopping a repository.
- Server: running one, its configuration, people, tokens and sign-in, the dashboard and the API.
- Pull requests: how much of a pull request agents wrote, and how those lines fared in review.
What is kept, and what is sent
- Data and privacy: the raw capture, the session record, what an upload contains and every file the binary keeps.
- Editions and license: what is Apache-2.0, and what is not.