If I try and implement advanced grouping strategies via the LLM/AI chat feature, I'm given a response that it's not possible with the tools it has available:
The blocker
The grouping-rules feature you documented (scope / run / filter / noRepeatWithin — the "strategy") is a per-channel strategy object. But none of the write tools I have (create_channel, update_channel) expose a strategy / grouping-rules parameter. The only schedule-shaping fields I can set are:
ordering (SHUFFLE / IN_ORDER / BY_AIR_DATE)
sortField / sortDir
There is no tool parameter to write the grouping strategy (the "Each show → Block of 2–4", "Movies → one at a time", noRepeatWithin, etc.) that your documentation describes. So I can't actually author these grouping rules through the available API — attempting to stuff them into filter would be wrong (filter is the content selector, not the group strategy).
How I'd like to proceed
I don't want to guess or fake it. A couple of options:
If there's a tool/parameter I'm missing to set a channel's grouping strategy (it may be surfaced differently — e.g. an update field name), tell me the field name and I'll apply the exact rule set below to every multi-show channel and re-generate.
The exact rule set I'd apply to each multi-show channel, ready to go the moment I can write it:
Rule 1 — Scope: Movies → Run: One at a time (1) (movies interspersed as single features)
Rule 2 — Scope: Each show → Run: Block of [2, 4] (random 2–4 episode mini-marathons per series)
Don't repeat a show: noRepeatWithin ≈ 3 shows (so you don't bounce back to the same series immediately)
Ordering left as SHUFFLE so series and movies rotate randomly.
(For channels of very short content — e.g. Classic Cartoons, some shorts — I'd instead use Run: ~Minutes {min:22, max:30} so 7-minute shorts pack into a proper block instead of just 2–4 tiny clips.)
Can you confirm the mechanism for writing the grouping strategy (option 1)? If update_channel accepts a strategy field under a name I haven't seen, share it and I'll roll this out across all the multi-show channels (and report the per-channel preview counts as I go). I'd rather confirm the write path than create something that silently doesn't take effect.
Which part of Airwave?
Server (admin + backend)
How is it installed / which platform?
Server — Docker / self-host image (ghcr.io/quixomatic/airwave)
Airwave version
0.12.37
What happened?
If I try and implement advanced grouping strategies via the LLM/AI chat feature, I'm given a response that it's not possible with the tools it has available:
Steps to reproduce
Logs / error output
Environment
No response
Before submitting