The API reports a timeout. The worker reports a retry. The authentication service has the first event in the sequence. When those events live in separate local files, following each file independently makes the sequence harder to inspect.
Flik 0.3.0 adds live multi-file merge and a three-column view. Here is a small, synthetic example you can reproduce without connecting a cloud account or sending logs to a service.
Download and verify
Download the archive for your platform and SHA256SUMS from Flik 0.3.0. Compare its SHA-256 before extracting, then run flik --version; it should print flik 0.3.0.
Create three local inputs
printf '%s\n' '2026-11-12T10:00:00Z INFO request=req-42 accepted' '2026-11-12T10:00:03Z ERROR request=req-42 upstream_timeout' > api.log printf '%s\n' '2026-11-12T10:00:01Z INFO request=req-42 job_started' '2026-11-12T10:00:02Z WARN request=req-42 retry_scheduled' > worker.log printf '%s\n' '2026-11-12T09:59:59Z INFO request=req-42 token_validated' > auth.log flik -m api.log worker.log auth.log Use a new empty directory for these synthetic fixtures. The existing events should begin with authentication at 09:59:59, followed by the API at 10:00:00, worker events at 10:00:01 and 10:00:02, and the API timeout at 10:00:03. Filename labels keep the source visible.
Follow an event that arrives later
In another terminal, in the same directory:
printf '%s\n' '2026-11-12T10:00:04Z INFO request=req-42 retry_completed' >> worker.log With follow mode enabled, the new worker event appears in the combined stream. The merger polls the local input files and appends to a temporary combined file. This differs from the existing in-app merge investigation command: CLI -m starts a stream that continues following file additions.
Keep three consecutive pages visible
Restart with:
flik -m -l api.log worker.log auth.log The columns are consecutive pages labelled p-2, p-1 and Current, read from left to right. They are not separate service columns. With only six fixture lines, older pages may be empty; append more synthetic events to explore the layout in a wide terminal. Tab or 4 cycles line, table and long-split modes. View changes and theme changes made with T are saved in Flik's configuration.
Know the boundaries
The initial inputs are loaded into memory, sorted by extracted timestamps, and written to a temporary file. Live events are sorted within each polling batch. An event received late is appended; it does not cause previous output to be globally reordered. This is not a distributed clock reconciliation mechanism.
The temporary combined file grows while following. Truncation is detected when file size decreases; replacing a file with one of equal or greater size can miss events. Use small representative inputs first, especially if your services rotate logs frequently. The byte-offset memory characteristics of ordinary single-file viewing do not describe the initial in-memory merge.
If you try this workflow, report a specific ordering or source-label issue with a small synthetic fixture, your OS and flik --version. Do not attach private production logs.


