"Be cautious about using FAAAH to process massive datasets. Some providers (you know which ones) might do some Kafkaesque interpretations of their ambiguous ToS, and deploy Orwellian telemetry to detect infractions (hasn't happened to me yet, YOLO!)" who do you mean by that
I just mean that some providers do some telemetry to detect (sometimes incorrectly) bad use. I think there were recent cases where OpenClaw-related strings in the prompt produced token exhaustion on Claude Code users.
I'm just saying that standard usage would produce a somewhat repetitive spawning of subagents which get prompts with a consistent formatting, and it's not entirely impossible that that plus large volume of use might trigger some detection somewhere.
I'm just saying that standard usage would produce a somewhat repetitive spawning of subagents which get prompts with a consistent formatting, and it's not entirely impossible that that plus large volume of use might trigger some detection somewhere.