HolobiomicsLab · lfnothias · May 5, 2026 · Apr 23, 2026 · Apr 23, 2026 · Apr 23, 2026
diff --git a/.github/workflows/pr_agent.yml b/.github/workflows/pr_agent.yml
@@ -6,9 +6,6 @@ on:
     branches:
       - main
       - dev
-      - dev_lfx
-      - dev_madina
-      - dev_Mkdocs
   workflow_dispatch:
         inputs:
           branch:

diff --git a/README.md b/README.md
@@ -23,6 +23,13 @@
 
 ---
 
+## v1.1.0 Highlights
+
+- Hardened the interpreter path by isolating LLM-generated Python behind trusted-mode controls and a subprocess runner instead of in-process execution.
+- Improved CLI safety with better file staging validation, collision handling, session-scoped logging, and consistent `--api-key` propagation through fallback SPARQL paths.
+- Made the Streamlit app easier to launch locally by stabilizing its import path handling and aligning the recommended startup command with the tested repo-root workflow.
+
+=======
 ## Demo
 
 Try the public MetaboT demonstrator at [metabot.holobiomicslab.eu](https://metabot.holobiomicslab.eu). It is connected to the Experimental Natural Products Knowledge Graph (ENPKG), an open metabolomics knowledge graph built from a chemodiverse collection of [1,600 plant extracts](https://doi.org/10.1093/gigascience/giac124).
@@ -99,6 +106,14 @@ conda env create -f environment.yml
 conda activate metabot
 ```
 
+To launch the application through Streamlit, install the dependencies and run the app from the repository root. In your terminal, execute:
+
+```bash
+pip install -r requirements.txt
+python -m streamlit run streamlit_webapp/streamlit_app.py
+```
+
+This repo-root launch path is the recommended setup for local development and matches the Streamlit smoke-tested workflow used in `v1.1.0`. You can provide your OpenAI key in the sidebar once the app starts, or preconfigure contributor/admin keys through environment variables if you use those deployment paths.
 If you prefer a plain virtual environment instead of Conda:
 
 ```bash

diff --git a/app/config/langgraph.json b/app/config/langgraph.json
@@ -3,14 +3,12 @@
         {"name": "Entry_Agent", "path": "app.core.agents.entry.agent", "llm_choice": "llm_o"},
         {"name": "ENPKG_agent", "path": "app.core.agents.enpkg.agent", "llm_choice": "llm_o"},
         {"name": "Sparql_query_runner", "path": "app.core.agents.sparql.agent", "llm_choice": "llm_o"},
-        {"name": "Interpreter_agent", "path": "app.core.agents.interpreter.agent", "llm_choice": "llm_o"},
         {"name": "supervisor", "path": "app.core.agents.supervisor.agent", "llm_choice": "llm_o"},
         {"name": "Validator", "path": "app.core.agents.validator.agent", "llm_choice": "llm_o"}
     ],
     "edges": [
         {"source": "ENPKG_agent", "target": "supervisor"},
-        {"source": "Sparql_query_runner", "target": "supervisor"},
-        {"source": "Interpreter_agent", "target": "supervisor"}
+        {"source": "Sparql_query_runner", "target": "supervisor"}
     ],
     "conditional_edges": [
         {
@@ -19,7 +17,6 @@
             "targets": [
                 {"condition_value": "ENPKG_agent", "target": "ENPKG_agent"},
                 {"condition_value": "Sparql_query_runner", "target": "Sparql_query_runner"},
-                {"condition_value": "Interpreter_agent", "target": "Interpreter_agent"},
                 {"condition_value": "FINISH", "target": "__end__"}
             ]
         },
@@ -41,6 +38,6 @@
     "entry_point": "Entry_Agent",
     "supervisor": {
         "name": "supervisor",
-        "members": ["ENPKG_agent", "Sparql_query_runner", "Interpreter_agent"]
+        "members": ["ENPKG_agent", "Sparql_query_runner"]
     }
 }
diff --git a/app/config/params.ini b/app/config/params.ini
@@ -1,16 +1,16 @@
 [llm_preview]
-id = gpt-4o
+id = gpt-5.4
 temperature = 0.3
 max_retries = 3
 
 [llm_o]
-id = gpt-4o
+id = gpt-5.4
 temperature = 0
 max_retries = 3
 
 
 [llm_mini]
-id = gpt-4o-mini
+id = gpt-5.4-mini
 temperature = 0
 max_retries = 3
 
@@ -48,7 +48,7 @@ max_retries = 3
 base_url = https://llama-3-1-70b-instruct.endpoints.kepler.ai.cloud.ovh.net/api/openai_compat/v1
 
 [llm_litellm_openai]
-id = gpt-4o
+id = gpt-5.4
 temperature = 0
 
 

diff --git a/app/core/agents/agents_factory.py b/app/core/agents/agents_factory.py
@@ -10,8 +10,6 @@
 
 logger = setup_logger(__name__)
 
-config = load_config()
-
 
 def create_all_agents(llms, graph, openai_key=None, session_id=None):
     """
@@ -27,6 +25,7 @@ def create_all_agents(llms, graph, openai_key=None, session_id=None):
         dict: A dictionary mapping agent names to their created executor instances.
     """
 
+    config = load_config()
     agents = config["agents"]
     executors = {}
 

diff --git a/app/core/agents/enpkg/tool_chemicals.py b/app/core/agents/enpkg/tool_chemicals.py
@@ -16,7 +16,7 @@
 from langchain_openai import OpenAIEmbeddings
 
 
-from langchain.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field
 from langchain.tools import BaseTool
 from typing import Optional
 
@@ -47,7 +47,7 @@ class ChemicalResolver(BaseTool):
         Dict[str, str]: a dictionary that contains the output chemical name and corresponding NPC Class URI.
     """
 
-    args_schema = ChemicalInput
+    args_schema: type[BaseModel] = ChemicalInput
     csv_data: List[Document] = None
     retriever: Any = None
     openai_key: str = None

diff --git a/app/core/agents/enpkg/tool_smiles.py b/app/core/agents/enpkg/tool_smiles.py
@@ -1,7 +1,7 @@
 import requests
 
 from langchain.tools import BaseTool
-from langchain.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field
 
 from typing import Optional
 
@@ -36,7 +36,7 @@ class SMILESResolver(BaseTool):
     smiles_string = "CCC12CCCN3C1C4(CC3)C(CC2)NC5=CC=CC=C45"
     inchikey = _run(smiles_string)
     """
-    args_schema = SMILESInput
+    args_schema: type[BaseModel] = SMILESInput
     openai_key: str = None
 
     def __init__(self, openai_key: str = None):

diff --git a/app/core/agents/enpkg/tool_target.py b/app/core/agents/enpkg/tool_target.py
@@ -5,7 +5,7 @@
 
 
 
-from langchain.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field
 
 from typing import Optional
 
@@ -38,7 +38,7 @@ class TargetResolver(BaseTool):
         str: A string containing the ChEMBLTarget notation.
     """
 
-    args_schema = TargetInput
+    args_schema: type[BaseModel] = TargetInput
     openai_key: str = None
 
     def __init__(self, openai_key: str = None):

diff --git a/app/core/agents/enpkg/tool_taxon.py b/app/core/agents/enpkg/tool_taxon.py
@@ -1,11 +1,9 @@
-from typing import Optional
+from typing import ClassVar, Optional
 
 from SPARQLWrapper import JSON, SPARQLWrapper
 
 
-from langchain.pydantic_v1 import BaseModel, Field
-
-from typing import Optional
+from pydantic import BaseModel, Field
 
 from langchain.callbacks.manager import (
     CallbackManagerForToolRun,
@@ -34,10 +32,10 @@ class TaxonResolver(BaseTool):
     Returns:
         str: A string that contains the Wikidata IRI if found, otherwise `None`.
     """
-    args_schema = TaxonInput
+    args_schema: type[BaseModel] = TaxonInput
 
-    ENDPOINT_URL = "https://query.wikidata.org/sparql"
-    PREFIXES = """
+    ENDPOINT_URL: ClassVar[str] = "https://query.wikidata.org/sparql"
+    PREFIXES: ClassVar[str] = """
         PREFIX prov: <http://www.w3.org/ns/prov#>
         PREFIX pr: <http://www.wikidata.org/prop/reference/>
         PREFIX wdt: <http://www.wikidata.org/prop/direct/>

diff --git a/app/core/agents/entry/tool_filesparser.py b/app/core/agents/entry/tool_filesparser.py
@@ -2,18 +2,22 @@
 import pandas as pd
 from datetime import datetime
 from langchain.tools import BaseTool
-from langchain.pydantic_v1 import BaseModel, Field
+from pydantic import BaseModel, Field
 from pathlib import Path
 from app.core.session import setup_logger, create_user_session
 
 logger = setup_logger(__name__)
 
+class FileAnalyzerInput(BaseModel):
+    """Input schema for FileAnalyzer - no input fields required."""
+    pass
+
 class FileAnalyzer(BaseTool):
     name: str = "FILE_ANALYZER"
     description: str = """
     Analyzes files in a specified directory and provides a summary of their content.
     """
-    args_schema = BaseModel  # Using BaseModel directly since no specific input fields are necessary
+    args_schema: type[BaseModel] = FileAnalyzerInput
     folder_path: Path = None
     openai_key: str = None
     session_id: str = None

diff --git a/app/core/agents/interpreter/agent.py b/app/core/agents/interpreter/agent.py
@@ -18,6 +18,7 @@ def create_agent(llms, graph, openai_key, session_id) -> AgentExecutor:
     tool_parameters = {
         "openai_key": openai_key,
         "session_id": session_id,
+        "llm_instance": llms.get(MODEL_CHOICE),
     }
 
     tools = import_tools(directory, module_prefix, **tool_parameters)

diff --git a/app/core/agents/interpreter/prompt.py b/app/core/agents/interpreter/prompt.py
@@ -23,9 +23,10 @@
 
     For using INTERPRETER_TOOL tool, you have to provide all the information needed for the interpretation of the question.This allows INTERPRETER_TOOL to analyze the information thoroughly and provide a clear, concise answer to the initial question.
     Your input for the tool should be a dictionary with one key, __arg1, and the value of this key should be a string contain at least all relevant complete filepaths, the user question and your demands for the tool. You should include, if any, the generated SPARQL query or other relevant information.
+    Wrap every filepath in quotes. This is mandatory for any filepath containing spaces and preferred for all filepaths.
     Your demand should explicitly say what you want from the tool. Please include 
     Example:
-        Filepath: Filepath provided by SPARQL_QUERY_RUNNER
+        Filepath: "Filepath provided by SPARQL_QUERY_RUNNER"
         User question: User question
         Demand: Interpret the file to give x and y information and provide a chart
         Other relevant information: Other information here