33 Commits

Author SHA1 Message Date
dongyuanjushi
4b752db8f0 feat: add environment with computer-use MCP (#500)
* add cua env

* update

* f

* update

* Update README

* Update README.md

* refactor action space

* update figure and readme

* fix async calling of osworld

* update instructions

* restructure action space

* update README

---------

Co-authored-by: aiosfoundation <aiosfoundation.org@gmail.com>
2025-06-11 01:53:01 -07:00
dongyuanjushi
920e35950f update model selection (#497)
* update

* update

* update

* update

* update

* update model config and selection

* u
2025-05-07 12:15:05 -04:00
dongyuanjushi
cbee47805b update (#495) 2025-04-30 13:36:32 -04:00
dongyuanjushi
1ce23679d9 refactor llm routing and update tests for routers (#491)
* update

* update

* update launch

* add more error messages for adapter

* fix smart routing

* update config

* update

* update test

* update test

* update test

* update workflow

* update

* update

* update

* update

* update

* update

* update

* update
2025-04-27 17:40:11 -04:00
dongyuanjushi
2b4452d0fe fix configuration refresh (#490) 2025-04-18 15:38:18 -04:00
dongyuanjushi
6083f6dd20 fix: sglang tool calling response format (#488)
* fix agent submit

* update scripts

* update process

* update scripts

* update routing strategy

* update scripts

* fix strategy

* update routing script

* update launching script

* fix decoding tool call

* fix tool calling issues for sglang backend

* update

* list available llms
2025-04-16 18:45:29 -04:00
dongyuanjushi
f09da0fea6 refactor launch script to load from config (#485)
* fix agent submit

* update scripts

* update process

* update scripts

* update routing strategy

* update scripts

* fix strategy

* update routing script

* update launching script
2025-04-03 11:17:12 -04:00
dongyuanjushi
1c6e001301 fix agent submit proxy (#482)
* fix agent submit

* update scripts

* update process

* update scripts

* update routing strategy

* update scripts

* fix strategy
2025-03-19 10:13:10 -04:00
WujiangXu
f3247bfdc6 inital commit 2025-03-10 13:51:33 -04:00
dongyuanjushi
8e47e1e1de fix context recover issue (#461)
* update litellm completion apis and fix launch of AIOS terminal

* update

* rm chroma local file

* update

* add more time for building kernel in github action

* add port for launching AIOS

* update config
2025-03-07 22:38:05 -05:00
Wujiang Xu
051d04bfa0 Amem0305 (#460)
* Initial commit for new branch

* memory

* fix bug

* add requirements

* add annotation
2025-03-06 04:27:46 -05:00
dongyuanjushi
e82528bfac update litellm completion apis and fix launch of AIOS terminal (#459) 2025-03-05 15:30:01 -05:00
dongyuanjushi
933dbffe65 fix context switch and logger (#454)
* fix the context switch issue

* add config example
2025-03-03 13:10:01 -05:00
dongyuanjushi
cfc9e650d0 refactor: update the import of cerebrum apis (#451)
* update README

* update

* refactor to migrate components setup to kernel.py

* rm breakpoint

* recover README

* update

* update

* update cerebrum apis

* update apis in the kernel
2025-02-26 17:05:14 -05:00
dongyuanjushi
e22bef6163 Refactor the yaml setup of kernel components (#450)
* update README

* update

* refactor to migrate components setup to kernel.py

* rm breakpoint

* recover README
2025-02-25 21:17:25 -05:00
dongyuanjushi
77debbc5ec feat: add file lock over file operations and fix redis issue (#447)
* fix collection creation issue and hash code for file ID

* fix redis connection issue

* update

* update instructions

* add warning of starting redis for rollback
2025-02-23 22:03:05 -05:00
dongyuanjushi
a3437f89b8 feat: integrate semantic filesystem (#436)
* integrate lsfs

* update dependencies

* update paper ref

* fix filesystem mount issue

* fix recursively scanning files issue and update README
2025-02-12 22:17:02 -05:00
xiang
c0f2437497 feat: Add dynamic LLM configuration support in AIOS kernel (#425)
* Fix

* Feat: Prioritize API Key from Config File

Problem: The system was using environment variables for API keys.
Solution: Changed the code to first check the config file for API keys. If not found, it falls back to environment variables.
Impact: Ensures API keys are managed centrally in the config file, improving flexibility and control.

* Update adapter.py

* fix: use PR code instead of main branch in workflow

Before: Workflow always used main branch code
After: Workflow uses PR code when testing PR

* Update test_ollama.yml

* Update test_ollama.yml

* Update test_ollama.yml

* Update README: Add AIOS Refresh Command

* feat: Add dynamic LLM configuration support in AIOS kernel

# Description
This PR enables AIOS kernel to dynamically handle LLM configurations from users, allowing:

1. Dynamic API key updates for different LLM providers
2. Runtime configuration refresh without restart
3. Secure API key management

## Changes
- Added config update endpoint in kernel
- Added configuration refresh mechanism
- Updated environment variable handling

* Update kernel.py
2025-01-17 08:24:02 -05:00
Joshua Park
997651caf4 feat: llm router (#393)
* chore: replaced endpoint-specific libraries with litellm

* feat!: litellm based llm router

* feat: implemented context manager for all llm endpoints

* feat: backwards compatibility for pre-router llm endpoints

* feat: implemented load balancing

* fix: added ollama workflow backwards compatibility endpoint

* chore: add documentation for creating load balancing strategies

* feat: use llm_backend for prefixing model names and create skeleton for local inference

* feat: implement vllm local backend

* feat: implement local huggingface

* feat: allow passing optional hostnames to router

* feat: part 1 of online huggingface and vllm impl

* feat: implement custom hostname inferencing of huggingface and vllm
2025-01-06 16:12:11 +08:00
dongyuanjushi
2a0fb0674e fix indent in kernel.py 2025-01-06 02:55:35 -05:00
Rick
e9660205e5 Fix incorrect agent name bug (#398)
Co-authored-by: Joshua Park <123624726+joshprk@users.noreply.github.com>
2025-01-06 15:31:36 +08:00
xiang
b79efd79c9 feat: Add Configuration Management and Refresh Option for AIOS Kernel (#391)
* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* Enhanced Error Message

dded better error logs in kernel.py to help find bugs faster.

## Changes Made

1. Added clear headers to show where we are in the code:
   - Tool Manager setup
   - Agent submission
   - Agent status checking

2. Made error messages better:
   - Clear error messages
   - Stack traces
   - Basic info like Agent ID

Now it's easier to:
- See what the code is doing
- Find where errors happen
- Fix problems quickly

* fix: improve agent loading and error handling

- Add debug logs to track agent loading
- Fix config handling in agent manager
- Update seeact_demo_agent config format

The code now:
- Shows better error messages
- Loads agents more reliably
- Keeps full config when packaging"

* Delete agent.py

* Add Configuration Management and Refresh Option for AIOS Kernel

Description:
Introduced a configuration file to manage global parameters for the AIOS kernel.
Updated the kernel to read parameters from this configuration file at launch.
Added a "refresh" command to reload and apply configuration changes without restarting the kernel.
Improved API key handling with secure masking in logs.
Enhanced error handling and logging for better debugging.

* Update .gitignore

* Update launch.py

* Update google.py

* implement refresh command

* Update config.yaml

* Update
2025-01-04 09:34:23 +08:00
xiang
00f84b140b fix: improve agent loading and error handling (#387)
* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* Enhanced Error Message

dded better error logs in kernel.py to help find bugs faster.

## Changes Made

1. Added clear headers to show where we are in the code:
   - Tool Manager setup
   - Agent submission
   - Agent status checking

2. Made error messages better:
   - Clear error messages
   - Stack traces
   - Basic info like Agent ID

Now it's easier to:
- See what the code is doing
- Find where errors happen
- Fix problems quickly

* fix: improve agent loading and error handling

- Add debug logs to track agent loading
- Fix config handling in agent manager
- Update seeact_demo_agent config format

The code now:
- Shows better error messages
- Loads agents more reliably
- Keeps full config when packaging"

* Delete agent.py
2024-12-30 09:42:16 +08:00
Balaji R
a6699e5f4f feat: add individual api key set up (#378) 2024-12-15 08:44:58 +08:00
xiang
8e690bec2a feat: Enhanced Error Logging (#372)
* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* feat: enhance API key documentation and display

* Enhanced Error Message

dded better error logs in kernel.py to help find bugs faster.

## Changes Made

1. Added clear headers to show where we are in the code:
   - Tool Manager setup
   - Agent submission
   - Agent status checking

2. Made error messages better:
   - Clear error messages
   - Stack traces
   - Basic info like Agent ID

Now it's easier to:
- See what the code is doing
- Find where errors happen
- Fix problems quickly
2024-12-09 05:53:14 +08:00
Kai Mei
7c2d809fc4 add cors for launching kernel (#371) 2024-12-08 10:05:19 -05:00
Kai Mei
25aa124072 Add exception messages (#370) 2024-12-07 09:05:24 -05:00
Joshua Park
71f5585f3c fix: remove hot reloading (#363) 2024-12-04 15:03:34 -05:00
Balaji R
1633094c2d V.20 (#342)
* changes

* changes

* changes

* tool db

* agenthhub routes

* server

* agenthub fix

* deploy

* linting

* linting deprecate

* linting deprecate

* code

* code

* tool loader demo

* Yawn

* tools

* code changes

* update

* f

* code

* code

* new deps

* code change

* update

* code changes

* changes

* update openai models

* update LLM callings

* update llm adapter

* merge main

* remove redundant files

* update installation

* rm deprecated actions

* update README

* update cuda requirements

* update README

* update python version

* update docs

* Update README.md

* update openai models

* update the README

* update readme

* update figure

* update README

* update README

* better adapter

* code

* update figure

* update figure

* fix response in local models

* update ollama

* fix tool calling issues in local models

* update figure

* fix backend issues and update docs

* update README

* fix the tool calling in openai models and update docs pointing to SDK

---------

Co-authored-by: dongyuanjushi <kai.mei@rutgers.edu>
2024-11-30 10:58:08 -05:00
BRama10
e121c5b0d1 feat: web ui connection with core AIOS (#208)
* refactor: exclude runtime to outer file

* feat: connection of web ui with aios

* feat: connection of web ui with aios

* bug: fresh exclude of cache
2024-08-10 13:43:10 -04:00
BRama10
da04e4394e feat: local runtime linked to AIOS (#202)
* chore: uvicorn solidary dep

* feat: local runtime linked to AIOS
2024-08-07 15:36:44 -07:00
Kai Mei
a6d05fe2a2 chore: add ruff necessity in PR template (#168) 2024-07-12 00:23:11 -04:00
BRama10
0737541fde feat: created AIOS runtime template (#164) 2024-07-09 12:42:25 -07:00