Sessions

ra persists every conversation as JSONL. Resume any session from any interface — CLI, REPL, or HTTP.

Resuming a session

CLI — resume the latest session:

ra --resume "Continue with the next step"

CLI — resume a specific session by ID:

ra --resume=<session-id> "Continue with the next step"

REPL — resume the latest session:

› /resume

REPL — resume a specific session by ID:

› /resume abc-123

HTTP API:

{
  "messages": [{ "role": "user", "content": "Continue" }],
  "sessionId": "abc-123"
}

Auto-save

Sessions are saved automatically after each turn. You never need to explicitly save.

To resume a specific session later (if newer sessions exist), use ra --resume=ses_abc123.

Listing sessions

Use the HTTP API or the Inspector to browse sessions:

curl http://localhost:3000/sessions

Configuration

Sessions are stored under {dataDir}/sessions/ (default: .ra/sessions/). Set dataDir to change the root data directory.

app:
  dataDir: .ra              # root for all runtime data
  storage:
    maxSessions: 100        # max sessions to keep (auto-prune oldest)
    ttlDays: 30             # auto-expire sessions older than this

Storage format

Sessions are stored as JSONL files — one line per message. This makes them easy to inspect, grep, or process with standard tools:

cat .ra/sessions/ses_abc123.jsonl | head -5

See also

• CLI — resuming sessions from the command line
• REPL — /resume and /clear commands
• HTTP API — sessionId field in requests
• Configuration — storage settings