!autochat Bot Acts Like a Real User #30

New issue

Closed

opened 2025-05-14 10:45:06 -04:00 by milo · 2 comments

milo commented

2025-05-14 10:45:06 -04:00

Owner

🔁 Updated Scope for `!autochat Bot Acts Like a Real User`

✅ Completed for MVP:

Modular autochat.py system
Passive reply trigger via on_message
Context-aware, persona-driven responses
.env-based toggles + cooldown control
Early prompt logic with behavior shaping

🔁 Refined Goal (Now):

“Enable Delta to participate passively in conversations where her presence is contextually relevant — without relying on scheduled posting or keyword spam.”

📝 Scope Adjustment

Feature	Status	Notes
Passive engagement via `on_message`	✅ Done	Already modular, live, and functional
Reply filtering via prompt	✅ MVP implemented	Needs tuning, but framework is solid
Scheduled participation (`scheduler/smart.py`)	🛑 Out of scope (for now)	Too complex for MVP, save for v2
Smart triggers (conflict, questions, etc.)	🔜 Nice-to-have	Could become `autochat.rules` module
Roasting on keyword triggers	🔜 Optional extension	Low-priority but fun
Fact-checking or Q&A hooks	❌ Post-MVP	Requires NLP / parsing, separate effort

🎯 New Working Title

Passive Autochat MVP (No Scheduler)
autochat-mvp: passive-only, context-aware, prompt-driven

## 🔁 Updated Scope for `!autochat Bot Acts Like a Real User` ### ✅ Completed for MVP: - Modular `autochat.py` system - Passive reply trigger via `on_message` - Context-aware, persona-driven responses - `.env`-based toggles + cooldown control - Early prompt logic with behavior shaping --- ### 🔁 Refined Goal (Now): > “Enable Delta to participate passively in conversations where her presence is contextually relevant — *without relying on scheduled posting or keyword spam.*” --- ### 📝 Scope Adjustment | Feature | Status | Notes | |--------|--------|-------| | Passive engagement via `on_message` | ✅ Done | Already modular, live, and functional | | Reply filtering via prompt | ✅ MVP implemented | Needs tuning, but framework is solid | | Scheduled participation (`scheduler/smart.py`) | 🛑 Out of scope (for now) | Too complex for MVP, save for v2 | | Smart triggers (conflict, questions, etc.) | 🔜 Nice-to-have | Could become `autochat.rules` module | | Roasting on keyword triggers | 🔜 Optional extension | Low-priority but fun | | Fact-checking or Q&A hooks | ❌ Post-MVP | Requires NLP / parsing, separate effort | --- ### 🎯 New Working Title **Passive Autochat MVP (No Scheduler)** `autochat-mvp: passive-only, context-aware, prompt-driven`

milo added this to the Alpha Build Ready milestone 2025-05-14 10:45:06 -04:00

milo added the

High Priority

💡feature

labels 2025-05-14 10:45:06 -04:00

milo self-assigned this 2025-05-14 10:45:06 -04:00

milo added this to the Polishing Phase project 2025-05-14 10:45:06 -04:00

milo referenced this issue

2025-05-14 10:47:18 -04:00

Bot Appears as User #9

milo referenced this issue

2025-05-14 11:16:24 -04:00

Bot Appears as User #9

milo referenced this issue from a commit

2025-06-01 16:13:08 -04:00

Added autochat #30. Not fully implemented but foundation is solid.

milo referenced this issue from a commit

2025-06-06 15:05:17 -04:00

Issue #30 Autochat. Is pretty much fully implemented. Further tweaks are needed but its good for now

milo commented

2025-06-06 15:08:35 -04:00

Author

Owner

Almost fully implemented. Not gonna resolve yet because engagement is needed to test.

milo commented

2025-06-07 19:49:51 -04:00

Author

Owner

Current State:

So the way this works is through triggers in the personality json file. When these triggers are used then the bot replies, it then has a decay function that looses interest gradually as the chat continues. This is far from perfect and will need tuning later on.

Improvements:

Using a multimodal system we can summarize messages and determine if the bot should respond given certain criteria, this can be compounded with knowledge bases as well as long and short memory function.

### Current State: So the way this works is through `triggers` in the personality `json` file. When these triggers are used then the bot replies, it then has a decay function that looses interest gradually as the chat continues. This is far from perfect and will need tuning later on. ### Improvements: Using a `multimodal` system we can summarize messages and determine if the bot should respond given certain criteria, this can be compounded with `knowledge bases` as well as long and short `memory` function.

milo closed this issue

2025-06-07 19:50:02 -04:00