Docs  /  Profiles & Agents batch menu

Profiles & Agents batch menu

Script: AI_Profiles_Agents_Menu.bat (repo root)

Interactive Windows menu for:

  1. MySQL application profiles (application_profiles in ai_settings)
  2. Agent fleet (agent registry / agent profile apply / agent start)
  3. Combined launches (load MySQL profile, then run agent commands)

Quick start

cd /d "E:\_AugmentedIntelligence\_src\_AugmentedIntelligence_v11.0 - Typed Commands"
AI_Profiles_Agents_Menu.bat

With overrides:

AI_Profiles_Agents_Menu.bat --host JUPITER --user root --password secret --exe "x64\Release-GPU-AVX2\_AugmentedIntelligence.exe"

Env vars: AI_EXE, AI_MYSQL_HOST, AI_MYSQL_USER, AI_MYSQL_PASSWORD.

The script auto-finds the first existing EXE under:

  • x64\Release-GPU-AVX2\
  • x64\Release-GPU-AVX512\
  • x64\Release-CPU-AVX2\
  • x64\Release-CPU-NoAVX\
  • installer\payload\Application\
# Action
1 MySQL profiles submenu
2 Agents submenu
3 Combined MySQL + agent launch recipes
4 Interactive REPL with a profile
5 Dear ImGui (--gui) with optional profile
6 Set MySQL host/user/password for this session
7 Pick EXE build path
8 Doctor: seed + list profiles + agent registry
0 Exit

MySQL profiles (submenu)

Uses CLI flags:

--list-profiles
--active-profile
--seed-profiles
--profile <name>
--save-profile <name>
--delete-profile <name>
--settings-mysql-primary
--mysql-host / --mysql-user / --mysql-password

Seeded names include (among others):
default, developer_build, gaming_rl, knowledge_rl, rsi_laboratory, legal_research, desktop_support, human_interaction, privacy_recording, multimodal_perception, offline_local, simulation_workshop, law_firm_sim, driving_lab.

Agents (submenu)

Uses one-shot:

_AugmentedIntelligence.exe ... --command "agent registry"
_AugmentedIntelligence.exe ... --command "agent profile apply multi_agent_orchestrator dry-run"
_AugmentedIntelligence.exe ... --command "agent start research_agent"

Fleet profiles (dry-run first in the menu):

Profile Purpose
multi_agent_orchestrator Human + desktop + training + safety
minecraft_rl_lab Minecraft RL + safety
neural_rl_lab RSI / training / remote LLM
help_desk_operations Desktop / app support
spreadsheet_simple_agents Workbook simple agents
ai_red_team_lab Security tabletop
safe_family_assistant Shadow family mode
engineering_lab EE / build-test

Also: agent emergency stop, agent stop all, simple agent registry.

Combined recipes

Examples:

  1. developer_build + agent profile apply multi_agent_orchestrator dry-run
  2. gaming_rl + minecraft_rl_lab dry-run
  3. Resident mode: --profile NAME --resident

Safety

  • Agent profile live apply and delete profile require a Y confirmation.
  • Prefer dry-run agent profiles until you trust the fleet.
  • Do not commit real MySQL passwords into the .bat; use session prompt or env.
  • Rebuild the EXE if commands fail with “unknown command”.
Generated from the project markdown docs on 2026-07-24. This is a static, self-contained site.