Behavioral Quality & UX
How the agent "feels" to interact with matters as much as what it can do. Response calibration, channel-appropriate formatting, knowing when to be proactive vs. quiet, and having real opinions without being a contrarian — these are the patterns that make an agent feel like a competent collaborator rather than a verbose chatbot.
This section covers behavioral patterns observed and refined across production operation on messaging surfaces (Telegram, Discord) and automated channels.
Key Problems
Effort-to-stakes mismatch
Quick question gets a 5-paragraph framework response. Complex request gets a one-liner. The agent struggles to match response depth to the actual stakes of the question. "What time is it?" doesn't need a methodology section.
Channel formatting is not one-size-fits-all
Telegram wants concise, no walls of text. Discord handles markdown but not tables well. Gists are for comprehensive, structured output. The agent needs different formatting instincts per channel — and the rules for this currently live in SOUL.md's "Response Discipline" section, which is easy to overlook.
Proactive engagement balance
Too proactive = noise. Not proactive enough = the agent feels passive and useless. The sweet spot depends on time of day (quiet hours), recent interaction frequency, and whether there's actually something worth saying.
Group chat behavior
In group contexts, the agent needs to know when to speak, when to react with an emoji, and when to stay completely silent. Over-participating in group chats is a fast way to become annoying.
Sycophancy
"Great question!" and "Absolutely!" before every response. Models default to agreement and praise. A useful agent has opinions, pushes back when the user is wrong, and doesn't pad responses with false enthusiasm.
What's Here
- Streaming & Queue Mode — Config tuning for responsive Discord streaming and mid-run message interruption via steer mode
Tracks
- Behavioral patterns — response calibration, effort matching, opinion formation
- Channel formatting rules — per-channel output conventions
- Proactive engagement heuristics — when to reach out, when to stay quiet
- Tone/voice calibration — sycophancy avoidance, authentic personality expression
Status
Initial pattern documented. The streaming/queue mode pattern is the first concrete writeup. More behavioral patterns need extraction from daily operational notes.