The run log contains nothing about what a submitted script did, so a Code Mode run can't be debugged from the console or the API.
Measured across 8 consecutive runs (5s–144s, from an A/B eval of Code Mode):
- Every run logs exactly 5 lines — 4 container/platform lines, then
{"ok":true}. Duration and workload make no difference.
- Script
stdout never reaches the log. One run emitted 9,286 bytes of stdout; none of it appears.
- Nested Actor runs are unlogged. One run started 40
apify/website-content-crawler runs; the log names none of them. Reconstructing what ran required apify runs ls per Actor.
- No signal while the script works.
statusMessage is null for the whole run, so nothing reports progress. A 286s run chaining ~12 nested Actor calls logged container start, then {"ok":true} 4m43s later.
Affected: debugging a Code Mode run from the console or API, and agents polling run status to find out what their script is doing.
Distinct from #2, which covers the terminal status of a failed script. This is the absence of execution detail in the log, which remains once #2 is fixed.
The run log contains nothing about what a submitted script did, so a Code Mode run can't be debugged from the console or the API.
Measured across 8 consecutive runs (5s–144s, from an A/B eval of Code Mode):
{"ok":true}. Duration and workload make no difference.stdoutnever reaches the log. One run emitted 9,286 bytes of stdout; none of it appears.apify/website-content-crawlerruns; the log names none of them. Reconstructing what ran requiredapify runs lsper Actor.statusMessageisnullfor the whole run, so nothing reports progress. A 286s run chaining ~12 nested Actor calls logged container start, then{"ok":true}4m43s later.Affected: debugging a Code Mode run from the console or API, and agents polling run status to find out what their script is doing.
Distinct from #2, which covers the terminal status of a failed script. This is the absence of execution detail in the log, which remains once #2 is fixed.