Loading README.md +5 −0 Changes for README.md: 5 added lines, 0 removed lines. Original line number Diff line number Diff line Loading @@ -21,6 +21,11 @@ AI Agent is a lightweight, modular backend built with **[`FastAPI`](https://fast It is designed to be easily extensible with support for different LLMs and OpenOP MCP servers. It keeps a short per-session chat memory in Redis when `REDIS_URL` is set. The LLM backend is selected with `LLM_PROVIDER`: `groq` (default, needs `GROQ_API_KEY`) or `openai` (any OpenAI-compatible endpoint such as Open WebUI; needs `OPENAI_BASE_URL`, `OPENAI_MODEL_NAME` and `OPENAI_API_KEY`). The model must support tool calling. See `ai_agent/README.md` for the settings. ## Authentication Every request carries the user's Keycloak access token (`Authorization: Bearer …`). The agent Loading ai_agent/README.md +21 −1 Changes for ai_agent/README.md: 21 added lines, 1 removed line. Original line number Diff line number Diff line Loading @@ -36,10 +36,19 @@ APP_PORT=9013 # --- MCP Server --- MCP_SERVER_URL=http://127.0.0.1:8004/mcp # --- LLM provider: groq (default) or openai --- LLM_PROVIDER=groq # --- Groq --- GROQ_API_KEY=your_groq_api_key_here GROQ_MODEL_NAME=qwen/qwen3.8-27b # --- OpenAI-compatible, e.g. Open WebUI (only when LLM_PROVIDER=openai) --- # OPENAI_BASE_URL=https://open-webui.example/api # OPENAI_MODEL_NAME=your-model # OPENAI_API_KEY=sk-... # OPENAI_TIMEOUT=120 # --- Optional short-term memory (Redis) --- REDIS_URL=redis://127.0.0.1:6379/0 MEMORY_WINDOW_TURNS=6 Loading @@ -47,6 +56,12 @@ MEMORY_TTL_SECONDS=3600 MEMORY_KEY_PREFIX=ai_agent:session ``` ### Using an OpenAI-compatible endpoint (for example Open WebUI) Set `LLM_PROVIDER=openai`, `OPENAI_BASE_URL`, `OPENAI_MODEL_NAME` and `OPENAI_API_KEY`. For Open WebUI the base URL is `https://<host>/api` and the key is created under Settings → Account → API Keys. The model must support tool calling. The endpoints keep their `/groq-mcp` paths. ### Environment Variables | Variable | Description | Loading @@ -54,8 +69,13 @@ MEMORY_KEY_PREFIX=ai_agent:session | `APP_HOST` | Host interface the server binds to | | `APP_PORT` | Port the backend server listens on | | `MCP_SERVER_URL` | URL of the MCP server used by sub-agent | | `GROQ_API_KEY` | API key for Groq LLM access | | `LLM_PROVIDER` | Model backend: `groq` (default) or `openai` | | `GROQ_API_KEY` | API key for Groq; required when `LLM_PROVIDER=groq` | | `GROQ_MODEL_NAME` | Groq model identifier | | `OPENAI_BASE_URL` | OpenAI-compatible endpoint (Open WebUI: `https://host/api`); required when `LLM_PROVIDER=openai` | | `OPENAI_MODEL_NAME` | Model identifier; required when `LLM_PROVIDER=openai` | | `OPENAI_API_KEY` | Bearer API key; required when `LLM_PROVIDER=openai` | | `OPENAI_TIMEOUT` | Request timeout in seconds (default `120`) | | `REDIS_URL` | Redis DSN for short-term memory (`redis://host:6379/0`) | | `MEMORY_WINDOW_TURNS` | Number of recent user/assistant turns to retain | | `MEMORY_TTL_SECONDS` | Session memory TTL in seconds | Loading Loading
README.md +5 −0 Changes for README.md: 5 added lines, 0 removed lines. Original line number Diff line number Diff line Loading @@ -21,6 +21,11 @@ AI Agent is a lightweight, modular backend built with **[`FastAPI`](https://fast It is designed to be easily extensible with support for different LLMs and OpenOP MCP servers. It keeps a short per-session chat memory in Redis when `REDIS_URL` is set. The LLM backend is selected with `LLM_PROVIDER`: `groq` (default, needs `GROQ_API_KEY`) or `openai` (any OpenAI-compatible endpoint such as Open WebUI; needs `OPENAI_BASE_URL`, `OPENAI_MODEL_NAME` and `OPENAI_API_KEY`). The model must support tool calling. See `ai_agent/README.md` for the settings. ## Authentication Every request carries the user's Keycloak access token (`Authorization: Bearer …`). The agent Loading
ai_agent/README.md +21 −1 Changes for ai_agent/README.md: 21 added lines, 1 removed line. Original line number Diff line number Diff line Loading @@ -36,10 +36,19 @@ APP_PORT=9013 # --- MCP Server --- MCP_SERVER_URL=http://127.0.0.1:8004/mcp # --- LLM provider: groq (default) or openai --- LLM_PROVIDER=groq # --- Groq --- GROQ_API_KEY=your_groq_api_key_here GROQ_MODEL_NAME=qwen/qwen3.8-27b # --- OpenAI-compatible, e.g. Open WebUI (only when LLM_PROVIDER=openai) --- # OPENAI_BASE_URL=https://open-webui.example/api # OPENAI_MODEL_NAME=your-model # OPENAI_API_KEY=sk-... # OPENAI_TIMEOUT=120 # --- Optional short-term memory (Redis) --- REDIS_URL=redis://127.0.0.1:6379/0 MEMORY_WINDOW_TURNS=6 Loading @@ -47,6 +56,12 @@ MEMORY_TTL_SECONDS=3600 MEMORY_KEY_PREFIX=ai_agent:session ``` ### Using an OpenAI-compatible endpoint (for example Open WebUI) Set `LLM_PROVIDER=openai`, `OPENAI_BASE_URL`, `OPENAI_MODEL_NAME` and `OPENAI_API_KEY`. For Open WebUI the base URL is `https://<host>/api` and the key is created under Settings → Account → API Keys. The model must support tool calling. The endpoints keep their `/groq-mcp` paths. ### Environment Variables | Variable | Description | Loading @@ -54,8 +69,13 @@ MEMORY_KEY_PREFIX=ai_agent:session | `APP_HOST` | Host interface the server binds to | | `APP_PORT` | Port the backend server listens on | | `MCP_SERVER_URL` | URL of the MCP server used by sub-agent | | `GROQ_API_KEY` | API key for Groq LLM access | | `LLM_PROVIDER` | Model backend: `groq` (default) or `openai` | | `GROQ_API_KEY` | API key for Groq; required when `LLM_PROVIDER=groq` | | `GROQ_MODEL_NAME` | Groq model identifier | | `OPENAI_BASE_URL` | OpenAI-compatible endpoint (Open WebUI: `https://host/api`); required when `LLM_PROVIDER=openai` | | `OPENAI_MODEL_NAME` | Model identifier; required when `LLM_PROVIDER=openai` | | `OPENAI_API_KEY` | Bearer API key; required when `LLM_PROVIDER=openai` | | `OPENAI_TIMEOUT` | Request timeout in seconds (default `120`) | | `REDIS_URL` | Redis DSN for short-term memory (`redis://host:6379/0`) | | `MEMORY_WINDOW_TURNS` | Number of recent user/assistant turns to retain | | `MEMORY_TTL_SECONDS` | Session memory TTL in seconds | Loading
