176 Commits

Author SHA1 Message Date
9d6a351ede 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 11:10:10 2026-03-24 11:10:10 +00:00
f7a49114cf 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 11:09:45 2026-03-24 11:09:45 +00:00
3437379df7 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 11:08:46 2026-03-24 11:08:46 +00:00
630f489789 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 11:00:27 2026-03-24 11:00:27 +00:00
d6b49e4fde 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:58:24 2026-03-24 10:58:24 +00:00
7f6f28fd0f 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:56:00 2026-03-24 10:56:00 +00:00
a31e2f7ff4 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:52:00 2026-03-24 10:52:00 +00:00
2e1f2530b8 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:48:28 2026-03-24 10:48:28 +00:00
bdd1cedbf7 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:42:30 2026-03-24 10:42:30 +00:00
bd4ce084a1 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:32:57 2026-03-24 10:32:57 +00:00
cbdbc07605 1 2026-03-24 10:32:41 +00:00
2d792355ec 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:23:43 2026-03-24 10:23:43 +00:00
28a5cd1417 🚀 Auto-deploy: BotVPS atualizado em 24/03/2026 10:18:32 2026-03-24 10:18:32 +00:00
82ca280628 24mar 2026-03-24 10:15:22 +00:00
38d0429b83 24mar26 2026-03-24 10:15:15 +00:00
2480846e31 refactor: Update credential synchronization to async Gitea fetch and simplify root HTML serving. 2026-03-24 02:26:23 +00:00
cb729809a9 Refactor: fix async errors, CRLF line endings and bot tokens. 2026-03-24 00:02:56 +00:00
b7e6239216 refatoracao 2026-03-23 23:38:56 +00:00
Marcos
8002262cf7 Change default Ollama model from qwen2.5-coder to llama3.2:1b for faster chat 2026-03-22 17:25:11 -03:00
Marcos
17dcb9d178 Increase Ollama timeout to 180s and add num_ctx 2026-03-22 16:51:21 -03:00
Marcos
2cc4ed0d18 Fix Ollama endpoint: use http://ollama:11434 2026-03-22 16:40:27 -03:00
Marcos
a74978da4a Add Ollama connection check and better error messages 2026-03-22 16:31:08 -03:00
Marcos
0084577a70 Add Gitea repo sync for credentials (admtracksteel/Keys) 2026-03-22 16:29:55 -03:00
Marcos
22af78e6a5 Fix /llm command to show status when no args 2026-03-22 16:24:06 -03:00
Marcos
62f90a381c Add segredos.md parser as fallback for credentials 2026-03-22 16:18:18 -03:00
Marcos
1ec63581c2 Fix format_completion_message counter and step display 2026-03-22 16:10:26 -03:00
Marcos
bd0cbf8769 fix: Correct Ollama endpoint and BotVPS path
- Ollama: Use ollama-lw4s8g4gc8gss4gkc4gg0wk4 hostname instead of localhost
- BotVPS: Path is /app inside container
- Improve detect_git_repo_path to find correct paths
- Update planner prompt with correct context
2026-03-22 16:03:41 -03:00
Marcos
9b429f5505 feat: Add credentials viewer with copy buttons
- Add 'Credenciais Carregadas' section in web interface
- Show services (Coolify, Gitea, Supabase, Logto) status
- Display path and number of keys found
- Add refresh button to reload credentials
- Add copy to clipboard functionality for credentials
2026-03-22 16:00:58 -03:00
Marcos
a229d1a64b fix: Use context.bot_data for plan persistence
- Store pending plan in context.bot_data (Telegram persists this)
- Add debug prints for troubleshooting
- Fixes plan not being retrieved on confirmation
2026-03-22 15:57:48 -03:00
Marcos
b782b611a2 fix: Save full plan for confirmation instead of just task
- Store complete plan in pending_plans dict
- When user confirms, execute stored plan directly
- Fixes 0/0 steps issue when confirming
2026-03-22 15:38:27 -03:00
Marcos
d7bc7df32c feat: Auto-sync credentials on startup
- Import credential_manager
- Call sync_credentials() on FastAPI startup event
- Print sync status in logs
2026-03-22 15:36:57 -03:00
Marcos
dc1a3f8bad feat: Add Orchestrator LLM config to web interface
- Add section to configure Planner (Gemini, OpenAI, Anthropic, Ollama)
- Add section to configure Executor (Ollama, Gemini)
- Auto-discover Ollama models
- Add Orchestrator status display
- Add Sync Credentials button
- Load models dynamically based on provider selection
2026-03-22 15:35:33 -03:00
Marcos
19ac2b55d4 feat: Auto-detect repository paths for any app
- Add detect_git_repo_path() function to find repos in /data/repositories
- Add detect_app_in_docker() to find containers by name
- Planner now receives detected context automatically
- Supports BotVPS, TrackSteel, and any repo in /data/repositories
- Uses 'docker compose' (space) syntax

Example: 'faz deploy do tracksteel' will detect TrackSteel repo path automatically
2026-03-22 15:30:15 -03:00
Marcos
7579e62d9b fix: Improve planner with correct paths and docker compose syntax
- Add BotVPS repository path: /data/applications/bw1erd9ww5121i1fsh420bcj
- Force use of 'docker compose' (space) not 'docker-compose' (hyphen)
- Use bash tool for all commands with full absolute paths
- Add clear examples of CORRECT vs INCORRECT commands
2026-03-22 15:27:59 -03:00
Marcos
0f7bdbc3a4 fix: Improve planner prompt for correct command generation
- Fix tool usage: command should be the ARGUMENT, not tool name
- Fix Docker Compose: use 'docker compose' (space) not 'docker-compose'
- Add clear examples for correct tool usage
- Improve command generation for git and docker operations
2026-03-22 15:25:51 -03:00
Marcos
9b45f7a393 fix: Docker format string escaping 2026-03-22 15:20:22 -03:00
Marcos
809578a9d6 feat: Integrate orchestrator with Telegram bot
- Add orchestrator imports and handlers
- Detect orchestrator tasks via keywords (deploy, docker, git, etc.)
- Add commands: /status, /tools, /sync, /orchestrate
- Add confirmation flow (sim/nao)
- Keep fallback to ai_agent for non-orchestrator tasks
2026-03-22 15:07:34 -03:00
Marcos
64731a24a5 Stability: CPU fix with psutil interval and LLM timeouts 2026-03-22 14:36:20 -03:00
Marcos
fde085835b Restore: Reverted ai_agent.py to last known stable commit (6cf2c30) 2026-03-22 14:05:17 -03:00
Marcos
39eeaf95bd Fix: Restored full agent capabilities and fixed server variables 2026-03-22 13:59:39 -03:00
Marcos
12427dcb46 Fix: Restored system_prompt_base and tools_desc to fix server crash (Error 500) 2026-03-22 13:57:17 -03:00
Marcos
4d153d7a9e Gemini: Corrected model to gemini-2.0-flash 2026-03-22 13:55:03 -03:00
Marcos
8b195b37fb Fix: Restored AI IQ and conversational fluency for local models 2026-03-22 13:23:23 -03:00
Marcos
a20b32b70b Performance: Implemented Turbo Lite mode for Ollama local 2026-03-22 13:16:39 -03:00
Marcos
03b1793be9 Optimization: Improved context handling and lowered Ollama temperature for better local performance 2026-03-22 13:08:52 -03:00
Marcos
3cfb6fb3d2 Fix: Robust web password handling for insight images 2026-03-22 12:58:11 -03:00
Marcos
029b8973b3 Fix: Final Web UI image rendering logic 2026-03-22 12:45:49 -03:00
Marcos
75d2a16fec Fix: Unified Context Memory between Web and Telegram 2026-03-22 12:40:06 -03:00
Marcos
6cf2c30e59 Fix: run_bash_command now returns stdout even on search errors 2026-03-22 12:23:49 -03:00
Marcos
8882b95650 Fix: AI Agent specialized on VPS_Sync path and image sending tags 2026-03-22 12:21:42 -03:00