chore: support new models

.gitignore

@@ -8,3 +8,24 @@ wheels/
 
 # Virtual environments
 .venv
+
+# Skills
+.codebuddy/
+.agent/
+.agents/
+.claude/
+.cursor/
+.gemini/
+.github/skills/
+.github/prompts/
+.opencode/
+.qoder/
+.qwen/
+.trae/
+
+# Environment variables
+.env
+.env.*
+*.env
+!.example.env
+!*.env.example

pyproject.toml

@@ -3,14 +3,14 @@ name = "llama-agent-skills"
 version = "0.1.0"
 description = "Add your description here"
 readme = "README.md"
-authors = [
+authors = []
-    { name = "Summer Shen", email = "3244742300@qq.com" }
-]
 requires-python = ">=3.11"
 dependencies = [
     "llama-index-core>=0.12.0",
-    "llama-index-llms-openai>=0.3.0",
+    "llama-index-llms-openai>=0.7.5",
     "python-frontmatter>=1.1.0",
+    "python-dotenv>=1.0.0",
+    "llama-index-llms-openai-like>=0.7.1",
 ]
 
 [project.scripts]

skills-lock.json

@@ -0,0 +1,10 @@
+{
+  "version": 1,
+  "skills": {
+    "excalidraw-diagram-generator": {
+      "source": "github/awesome-copilot",
+      "sourceType": "github",
+      "computedHash": "e24e3c4cab4b5a07bc1d1801db14868f3a1cd184f7f28f5dd2c4a2b234bf2f2d"
+    }
+  }
+}

src/llama_agent_skills/__init__.py

@@ -5,6 +5,8 @@ import asyncio
 import sys
 from pathlib import Path
 
+from dotenv import load_dotenv
+
 from llama_agent_skills.agent import create_agent, run_agent
 from llama_agent_skills.config import Config
 from llama_agent_skills.registry import SkillRegistry
@@ -29,6 +31,7 @@ async def _run_interactive(config: Config, registry: SkillRegistry) -> None:
         skills=skills,
         model=config.llm_model,
         api_key=config.llm_api_key,
+        base_url=config.llm_base_url or None,
         base_system_prompt=config.base_system_prompt,
     )
 
@@ -55,11 +58,13 @@ async def _run_interactive(config: Config, registry: SkillRegistry) -> None:
 
 def main() -> None:
     args = _parse_args()
+    load_dotenv()
     config = Config.from_env()
 
     if args.model:
         config = Config(
             llm_api_key=config.llm_api_key,
+            llm_base_url=config.llm_base_url,
             llm_model=args.model,
             skill_directories=config.skill_directories,
             base_system_prompt=config.base_system_prompt,

src/llama_agent_skills/agent.py

@@ -2,11 +2,60 @@ from __future__ import annotations
 
 from typing import Any, Callable
 
+import tiktoken
 from llama_index.core.agent.workflow import AgentWorkflow
+from llama_index.core.base.llms.types import LLMMetadata
 from llama_index.llms.openai import OpenAI
+from llama_index.llms.openai.utils import (
+    CHAT_MODELS,
+    is_chat_model,
+    is_function_calling_model,
+    openai_modelname_to_contextsize,
+)
 
 from llama_agent_skills.skill import Skill
 
+_DEFAULT_CONTEXT_WINDOW = 131072
+_FALLBACK_ENCODING = "cl100k_base"
+
+
+class _FlexibleOpenAI(OpenAI):
+    """OpenAI subclass that tolerates unknown model names for custom endpoints."""
+
+    _custom_context_window: int = _DEFAULT_CONTEXT_WINDOW
+
+    @property
+    def metadata(self) -> LLMMetadata:
+        model_name = self._get_model_name()
+        known_model = model_name in CHAT_MODELS
+
+        try:
+            context_window = openai_modelname_to_contextsize(model_name)
+        except ValueError:
+            context_window = self._custom_context_window
+
+        if known_model:
+            chat = is_chat_model(model=model_name)
+            func_calling = is_function_calling_model(model=model_name)
+        else:
+            chat = True
+            func_calling = True
+
+        return LLMMetadata(
+            context_window=context_window,
+            num_output=self.max_tokens or -1,
+            is_chat_model=chat,
+            is_function_calling_model=func_calling,
+            model_name=self.model,
+        )
+
+    @property
+    def _tokenizer(self) -> tiktoken.Encoding | None:
+        try:
+            return tiktoken.encoding_for_model(self._get_model_name())
+        except KeyError:
+            return tiktoken.get_encoding(_FALLBACK_ENCODING)
+
 
 def build_system_prompt(base_prompt: str, skills: list[Skill]) -> str:
     if not skills:
@@ -28,6 +77,7 @@ def create_agent(
     tools: list[Callable[..., Any]] | None = None,
     model: str = "gpt-4o-mini",
     api_key: str | None = None,
+    base_url: str | None = None,
     base_system_prompt: str = "You are a helpful assistant.",
 ) -> AgentWorkflow:
     system_prompt = build_system_prompt(base_system_prompt, skills or [])
@@ -35,10 +85,12 @@ def create_agent(
     llm_kwargs: dict[str, Any] = {"model": model}
     if api_key:
         llm_kwargs["api_key"] = api_key
+    if base_url:
+        llm_kwargs["api_base"] = base_url
 
     return AgentWorkflow.from_tools_or_functions(
         tools_or_functions=tools or [],
-        llm=OpenAI(**llm_kwargs),
+        llm=_FlexibleOpenAI(**llm_kwargs),
         system_prompt=system_prompt,
     )
 

src/llama_agent_skills/config.py

@@ -8,6 +8,7 @@ from pathlib import Path
 @dataclass(frozen=True)
 class Config:
     llm_api_key: str = ""
+    llm_base_url: str = ""
     llm_model: str = "gpt-4o-mini"
     skill_directories: list[Path] = field(default_factory=list)
     base_system_prompt: str = "You are a helpful assistant."
@@ -16,17 +17,24 @@ class Config:
     def from_env(cls) -> Config:
         """Create Config from environment variables.
 
-        OPENAI_API_KEY, LLM_MODEL, SKILL_DIRECTORIES (comma-separated paths),
+        OPENAI_API_KEY, OPENAI_BASE_URL, LLM_MODEL / LLM_MODEL_NAME,
-        BASE_SYSTEM_PROMPT.
+        SKILL_DIRECTORIES (comma-separated paths), BASE_SYSTEM_PROMPT.
         """
         skill_dirs_raw = os.environ.get("SKILL_DIRECTORIES", "")
         skill_directories = [
             Path(d.strip()) for d in skill_dirs_raw.split(",") if d.strip()
         ]
 
+        model = (
+            os.environ.get("LLM_MODEL")
+            or os.environ.get("LLM_MODEL_NAME")
+            or "gpt-4o-mini"
+        )
+
         return cls(
             llm_api_key=os.environ.get("OPENAI_API_KEY", ""),
-            llm_model=os.environ.get("LLM_MODEL", "gpt-4o-mini"),
+            llm_base_url=os.environ.get("OPENAI_BASE_URL", ""),
+            llm_model=model,
             skill_directories=skill_directories,
             base_system_prompt=os.environ.get(
                 "BASE_SYSTEM_PROMPT", "You are a helpful assistant."

uv.lock

@@ -714,13 +714,17 @@ source = { editable = "." }
 dependencies = [
     { name = "llama-index-core" },
     { name = "llama-index-llms-openai" },
+    { name = "llama-index-llms-openai-like" },
+    { name = "python-dotenv" },
     { name = "python-frontmatter" },
 ]
 
 [package.metadata]
 requires-dist = [
     { name = "llama-index-core", specifier = ">=0.12.0" },
-    { name = "llama-index-llms-openai", specifier = ">=0.3.0" },
+    { name = "llama-index-llms-openai", specifier = ">=0.7.5" },
+    { name = "llama-index-llms-openai-like", specifier = ">=0.7.1" },
+    { name = "python-dotenv", specifier = ">=1.0.0" },
     { name = "python-frontmatter", specifier = ">=1.1.0" },
 ]
 
@@ -778,15 +782,28 @@ wheels = [
 
 [[package]]
 name = "llama-index-llms-openai"
-version = "0.7.4"
+version = "0.7.5"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "llama-index-core" },
     { name = "openai" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/40/15/1c79fd19e19c38abcb4d6f16bc4c7ce36651bef1efa31973e01603df235b/llama_index_llms_openai-0.7.4.tar.gz", hash = "sha256:e7e078816babe33b29fa17ea0a771cf9589b2e95f1902cd8f1ec283a3a35c675", size = 27323, upload-time = "2026-03-27T15:32:18.154Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/65/27/18a7fd0873023aed145332dab5a09b95b298e4fff1c21685eaf22b629d87/llama_index_llms_openai-0.7.5.tar.gz", hash = "sha256:54123e679a7cddc1f2e969f278a4654050730daf84691731a0c53ae14feac3c7", size = 27423, upload-time = "2026-03-30T16:30:33.973Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/50/e8/1edce9dad3c7b55e37ca27723f7e8918e71e270bb3e67b49e90abdfaa32f/llama_index_llms_openai-0.7.4-py3-none-any.whl", hash = "sha256:b2cf35ce0d0ecf3209a0406da7383efdbd8f70a1c83eef287e4346c36ea24cba", size = 28378, upload-time = "2026-03-27T15:32:17.205Z" },
+    { url = "https://files.pythonhosted.org/packages/63/62/a847e9a94c2f92926c30188259f9f86e019dcc45122bbb222dea35a74c02/llama_index_llms_openai-0.7.5-py3-none-any.whl", hash = "sha256:c302c6386873420df3714c3d538f45379b6de27ab6a531f30c67419b39a538f5", size = 28492, upload-time = "2026-03-30T16:30:32.979Z" },
+]
+
+[[package]]
+name = "llama-index-llms-openai-like"
+version = "0.7.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "llama-index-core" },
+    { name = "llama-index-llms-openai" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/9d/9f/0d98d022a08f43d4374998072636ec50a7cb50009bbb9a2761f5b26a78cc/llama_index_llms_openai_like-0.7.1.tar.gz", hash = "sha256:ce7cef3686b1e62d7c08134b4d8ca56706cca816e4c4098eaede33002829a6f9", size = 5177, upload-time = "2026-03-13T16:15:58.156Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/6d/22/a1e1ec1112c69ca0a379cd72691c36cdbcba78362622ce9a27e5a97965cc/llama_index_llms_openai_like-0.7.1-py3-none-any.whl", hash = "sha256:831f1144077c6f9ea7a62e987b7f2af00310dded3056edca2cb509f70a3e650a", size = 4860, upload-time = "2026-03-13T16:15:59.113Z" },
 ]
 
 [[package]]
@@ -1462,6 +1479,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/36/c7/cfc8e811f061c841d7990b0201912c3556bfeb99cdcb7ed24adc8d6f8704/pydantic_core-2.41.5-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:56121965f7a4dc965bff783d70b907ddf3d57f6eba29b6d2e5dabfaf07799c51", size = 2145302, upload-time = "2025-11-04T13:43:46.64Z" },
 ]
 
+[[package]]
+name = "python-dotenv"
+version = "1.2.2"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/82/ed/0301aeeac3e5353ef3d94b6ec08bbcabd04a72018415dcb29e588514bba8/python_dotenv-1.2.2.tar.gz", hash = "sha256:2c371a91fbd7ba082c2c1dc1f8bf89ca22564a087c2c287cd9b662adde799cf3", size = 50135, upload-time = "2026-03-01T16:00:26.196Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/0b/d7/1959b9648791274998a9c3526f6d0ec8fd2233e4d4acce81bbae76b44b2a/python_dotenv-1.2.2-py3-none-any.whl", hash = "sha256:1d8214789a24de455a8b8bd8ae6fe3c6b69a5e3d64aa8a8e5d68e694bbcb285a", size = 22101, upload-time = "2026-03-01T16:00:25.09Z" },
+]
+
 [[package]]
 name = "python-frontmatter"
 version = "1.1.0"