mirror of
https://github.com/ParisNeo/lollms_hub.git
synced 2026-05-04 03:01:01 -04:00
2a5e6b4cd6
This commit updates several core components related to administration, proxy handling, bot management, and application startup sequence. **Changes include:** * **`app/api/v1/routes/admin.py`**: Added handling for Redis password updates in the admin settings endpoint. * **`app/api/v1/routes/proxy.py`**: Updated the shared vectorizer function to cache model description embeddings for performance. * **`app/core/bot_manager.py`**: Refined the logic for generating unique request IDs for bots. * **`app/main.py`**: Adjusted the application lifespan hook to ensure necessary directories are created during startup.
40 lines
1.7 KiB
Python
40 lines
1.7 KiB
Python
language:python
|
|
TOOL_LIBRARY_NAME = 'Skill Architect Loader'
|
|
TOOL_LIBRARY_DESC = 'Dynamically search and load LoLLMs SKILL.md files into context.'
|
|
TOOL_LIBRARY_ICON = '📜'
|
|
|
|
def init_tool_library() -> None:
|
|
import pipmaster as pm
|
|
pm.ensure_packages(["numpy", "safe-store"])
|
|
|
|
async def tool_load_specialized_skill(args: dict, lollms=None):
|
|
'''
|
|
Search the Hub library for a specialized skill or workflow and retrieve its instructions.
|
|
Use this when you encounter a task you don't have a specific protocol for.
|
|
|
|
Args:
|
|
args: dict with keys:
|
|
- query (str): The functional requirement (e.g. 'code review', 'legal analysis')
|
|
- search_mode (str, optional): 'vector' for semantic or 'keyword' for exact matching. Default 'vector'.
|
|
'''
|
|
from app.core.skills_manager import SkillsManager
|
|
import asyncio
|
|
|
|
query = args.get('query')
|
|
mode = args.get('search_mode', 'vector')
|
|
|
|
# We attempt to use the Hub's vectorizer if provided by the host interface
|
|
# In the Hub, the vectorizer is accessible via app state which isn't directly
|
|
# in the tool process, so we use the SkillsManager's internal logic.
|
|
|
|
# Note: LollmsSystem doesn't expose the vectorizer yet, so we fall back to keyword
|
|
# unless the tool initializes its own ST model (heavy).
|
|
# REPAIR: We'll perform a direct search.
|
|
|
|
results = await SkillsManager.search_skills(query, limit=1)
|
|
|
|
if not results:
|
|
return f"No specialized skills found for '{query}'."
|
|
|
|
skill = results[0]
|
|
return f"--- LOADED SKILL: {skill['name']} ---\n\n{skill['raw']}\n\n### INSTRUCTION: Incorporate the logic above into your next steps." |