vamplabAI · EvilFreelancer · Jul 10, 2026 · Mar 29, 2026 · Mar 29, 2026 · Mar 29, 2026
diff --git a/README.md b/README.md
@@ -126,6 +126,23 @@ The `sgrsh` command:
 
 For more examples and detailed usage instructions, see the [examples/](examples/) directory.
 
+### Agent Client Protocol (`sgracp`)
+
+Editors and tools that speak the [Agent Client Protocol](https://agentclientprotocol.com/) can run SGR agents over **stdio** (newline-delimited JSON-RPC), using the same YAML configuration as the HTTP server.
+
+```bash
+sgracp --config examples/sgr_deep_research/config.yaml
+```
+
+Optional `acp` block in `config.yaml` selects which `agents:` entry to expose (if omitted, the first agent definition is used):
+
+```yaml
+acp:
+  agent: sgr_agent
+```
+
+Protocol details follow the official ACP specification, the [Python SDK](https://agentclientprotocol.github.io/python-sdk/), and the `agent-client-protocol` package used by this binary.
+
 ## Benchmarking
 
 ![SimpleQA Benchmark Comparison](https://github.com/vamplabAI/sgr-agent-core/blob/main/docs/assets/images/simpleqa_benchmark_comparison.png)
@@ -156,7 +173,7 @@ If you have any questions - feel free to join our [community chat](https://t.me/
 
 This project is developed by the **neuraldeep** community. It is inspired by the Schema-Guided Reasoning (SGR) work and [SGR Agent Demo](https://abdullin.com/schema-guided-reasoning/demo)↗️ delivered by "LLM Under the Hood" community and AI R&D Hub of [TIMETOACT GROUP Österreich](https://www.timetoact-group.at)↗️
 
-![](./docs/assets/images/rmr750x200.png)
+![](rmr750x200.png)
 
 This project is supported by the AI R&D team at red_mad_robot, providing research capacity, engineering expertise, infrastructure, and operational support.
 

diff --git a/config.yaml.example b/config.yaml.example
@@ -27,6 +27,13 @@ execution:
   logs_dir: "logs"  # Directory for saving agent execution logs
   reports_dir: "reports"  # Directory for saving agent reports
 
+# Optional: Agent Client Protocol (stdio) via `sgracp --config config.yaml`
+# acp:
+#   agent: "sgr_agent"       # Default agent; must match a name under agents:
+#   models:                  # Extra models offered in the ACP "model" selector,
+#     - "gpt-4o"             # in addition to each agent's own configured model.
+#     - "o3-mini"            # Clients let the user switch agent and model at runtime.
+
 # Prompts Configuration
 # prompts:
 #   # Option 1: Use file paths (absolute or relative to project root)

diff --git a/docs/en/framework/agents.md b/docs/en/framework/agents.md
@@ -3,7 +3,7 @@
 This section dives into the technical implementation of agents and explores how to adapt them to your needs.
 
 !!! Tip "Agent Architecture"
-    Information about available agents and their differences can be found in [Main Concepts](main-concepts.md#agent).
+    Information about available agents and their differences can be found in [Main Concepts](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/main-concepts.md#agent).
 
 
 *For a detailed understanding of the full logic, it's better to familiarize yourself with the [source code](https://github.com/vamplabAI/sgr-agent-core/blob/main/sgr_agent_core/base_agent.py).*
@@ -137,7 +137,7 @@ When creating custom solutions, pay attention first and foremost to these method
 Stores all agent settings: LLM parameters, search, execution, prompts, and MCP configuration.
 
 !!! Tip "More about configuration"
-    Full description of the configuration system, settings hierarchy, and usage examples can be found in the [configuration guide](configuration.md).
+    Full description of the configuration system, settings hierarchy, and usage examples can be found in the [configuration guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md).
 
 #### Extending Configuration
 

diff --git a/docs/en/framework/configuration.md b/docs/en/framework/configuration.md
@@ -67,6 +67,36 @@ config = GlobalConfig.from_yaml("config.yaml")
 
 An example can be found in [`config.yaml.example`](https://github.com/vamplabAI/sgr-agent-core/blob/main/config.yaml.example).
 
+### Agent Client Protocol (`sgracp`)
+
+The optional top-level `acp` section configures the [`sgracp`](https://github.com/vamplabAI/sgr-agent-core) stdio entrypoint (Agent Client Protocol over stdin/stdout). It uses the same `agents:` definitions as the rest of the stack.
+
+| Field | Description |
+| ----- | ----------- |
+| `agent` | Name of an entry under `agents:` used as the **default** when a client connects. If omitted, the first agent in `agents:` is used. |
+| `models` | Optional list of extra model ids offered in the ACP model selector, in addition to the model each agent already declares. |
+
+Example:
+
+```yaml
+acp:
+  agent: sgr_agent
+  models:
+    - gpt-4o
+    - o3-mini
+```
+
+You can also set `SGR__ACP__AGENT` in the environment (see `pydantic-settings` nested env rules for your version).
+
+#### Switching agent and model at runtime
+
+The bridge exposes the active **agent** and **model** as [ACP session config options](https://agentclientprotocol.com/protocol/v1/session-config-options), so compatible clients (Zed, etc.) let the user switch them from the UI during a session:
+
+- The **Agent** selector lists every entry under `agents:`; the chosen one is used for the next turn.
+- The **Model** selector lists each agent's configured model plus anything in `acp.models`; the chosen model overrides `llm.model` for that session only (the agent definition is left untouched).
+
+Changes apply to the next turn; an in-flight turn keeps its current settings. This replaces the removed `session/set_model` method — model selection now goes through `session/set_config_option`.
+
 ### Observability and Langfuse integration
 
 SGR Agent Core supports optional [Langfuse](https://langfuse.com) integration for LLM tracing:

diff --git a/docs/en/framework/first-steps.md b/docs/en/framework/first-steps.md
@@ -56,7 +56,7 @@ if __name__ == "__main__":
 1. This agent uses exclusively native tool calling, making it well-suited for getting started
 2. Default parameters will be set here if you don't override some of them
 3. For the example, we'll take a minimal pair of tools that will allow getting an answer to the task and completing the agent's work.
-   Learn more about tools in the [tools documentation](tools.md)
+   Learn more about tools in the [tools documentation](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools.md)
 
 ???example "Example of Agent Reasoning Log"
     ```log

diff --git a/docs/en/framework/main-concepts.md b/docs/en/framework/main-concepts.md
@@ -2,7 +2,7 @@
 
 This document describes the key entities of the SGR Deep Research Framework.
 
-![SGR Agent Core Concept](../../assets/images/sgr_concept.png)
+![SGR Agent Core Concept](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/assets/images/sgr_concept.png)
 
 ## Vibe Glossary
 
@@ -156,7 +156,7 @@ All created tools are automatically registered in `ToolRegistry`
 **MCPBaseTool** — Base class for tools integrated with MCP (Model Context Protocol) servers. Handles calls through MCP client, converts them to the framework tool format
 
 !!!tip
-    *For more details on tools and their usage: [Tools](tools.md)*
+    *For more details on tools and their usage: [Tools](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools.md)*
 
 ## Definition
 
@@ -170,7 +170,7 @@ Agent object - Agent+Definition: Ready agent with isolated context, history, log
 
 **AgentConfig** — centralized agent configuration combining LLM, search, execution, prompts, and MCP settings. Supports hierarchical configuration system through `GlobalConfig` and `AgentDefinition` with automatic parameter inheritance and override.
 !!!tip
-    *For more details on configuration and definitions: [Configuration Guide](configuration.md)*
+    *For more details on configuration and definitions: [Configuration Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md)*
 
 
 ## Registry

diff --git a/docs/en/framework/qna.md b/docs/en/framework/qna.md
@@ -21,7 +21,7 @@
 
 **4. Wrapper and infrastructure**
 
-- **`AgentDefinition` and `AgentConfig`** — the [documentation](configuration.md) already explains why it's so complex
+- **`AgentDefinition` and `AgentConfig`** — the [documentation](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md) already explains why it's so complex
 - **`services/registry` + `agent_factory`** — the final picture of how everything works will come together
 
 **5. Remaining topics**

diff --git a/docs/en/framework/tools.md b/docs/en/framework/tools.md
@@ -10,15 +10,15 @@ Tools are divided into two categories:
 
 | Item | Category | Description |
 | --- | --- | --- |
-| [ReasoningTool](tools/reasoning-tool.md) | System | Core tool for Schema-Guided Reasoning agents that determines the next reasoning step |
-| [FinalAnswerTool](tools/final-answer-tool.md) | System | Final answer tool that completes the research task and updates agent state |
-| [CreateReportTool](tools/create-report-tool.md) | System | Tool for generating a detailed research report with inline citations and saving it to disk |
-| [ClarificationTool](tools/clarification-tool.md) | System | Tool for asking clarification questions and pausing execution until user response |
-| [GeneratePlanTool](tools/generate-plan-tool.md) | System | Tool for creating an initial research plan and breaking a request into steps |
-| [AdaptPlanTool](tools/adapt-plan-tool.md) | System | Tool for updating an existing research plan based on new information |
-| [WebSearchTool](tools/web-search-tool.md) | Auxiliary | Web search tool powered by Tavily Search API for fresh information |
-| [ExtractPageContentTool](tools/extract-page-content-tool.md) | Auxiliary | Tool for extracting full content from specific web pages using Tavily Extract API |
-| [RunCommandTool](tools/run-command.md) | Auxiliary | Tool for executing shell commands in safe or unsafe mode inside a workspace boundary |
+| [ReasoningTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/reasoning-tool.md) | System | Core tool for Schema-Guided Reasoning agents that determines the next reasoning step |
+| [FinalAnswerTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/final-answer-tool.md) | System | Final answer tool that completes the research task and updates agent state |
+| [CreateReportTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/create-report-tool.md) | System | Tool for generating a detailed research report with inline citations and saving it to disk |
+| [ClarificationTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/clarification-tool.md) | System | Tool for asking clarification questions and pausing execution until user response |
+| [GeneratePlanTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/generate-plan-tool.md) | System | Tool for creating an initial research plan and breaking a request into steps |
+| [AdaptPlanTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/adapt-plan-tool.md) | System | Tool for updating an existing research plan based on new information |
+| [WebSearchTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/web-search-tool.md) | Auxiliary | Web search tool powered by Tavily Search API for fresh information |
+| [ExtractPageContentTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/extract-page-content-tool.md) | Auxiliary | Tool for extracting full content from specific web pages using Tavily Extract API |
+| [RunCommandTool](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/tools/run-command.md) | Auxiliary | Tool for executing shell commands in safe or unsafe mode inside a workspace boundary |
 
 ## BaseTool
 

diff --git a/docs/en/getting-started/index.md b/docs/en/getting-started/index.md
@@ -55,7 +55,7 @@ If you want to use SGR Agent Core as a Python library (framework):
 pip install sgr-agent-core
 ```
 
-See the [Installation Guide](installation.md) for detailed instructions and the [Using as Library](../framework/first-steps.md) guide to get started.
+See the [Installation Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/getting-started/installation.md) for detailed instructions and the [Using as Library](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/first-steps.md) guide to get started.
 
 ### CLI Tool (`sgrsh`)
 
@@ -115,19 +115,19 @@ if __name__ == "__main__":
 
 ## Documentation
 
-- **[Highlights](../highlights.md)** — What's cool about this framework: short cases and intuitive examples
-- **[Installation](installation.md)** — Detailed installation instructions for pip and Docker
-- **[Agent Core Framework](../framework/main-concepts.md)** — Understand the core concepts and architecture
-- **[Using as Library](../framework/first-steps.md)** — Learn how to use SGR Agent Core as a Python library
-- **[API Server Quick Start](../sgr-api/SGR-Quick-Start.md)** — Get started with the REST API service
+- **[Highlights](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/highlights.md)** — What's cool about this framework: short cases and intuitive examples
+- **[Installation](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/getting-started/installation.md)** — Detailed installation instructions for pip and Docker
+- **[Agent Core Framework](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/main-concepts.md)** — Understand the core concepts and architecture
+- **[Using as Library](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/first-steps.md)** — Learn how to use SGR Agent Core as a Python library
+- **[API Server Quick Start](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/sgr-api/SGR-Quick-Start.md)** — Get started with the REST API service
 
 ## Contact & Community
 
 **For collaboration inquiries**: [@VaKovaLskii](https://t.me/neuraldeep)
 
 **Community Chat**: We answer questions in [Telegram chat](https://t.me/sgragentcore) (ru/en)
 
-![](../../assets/images/rmr750x200.png)
+![](rmr750x200.png)
 
 This project is supported by the AI R&D team at red_mad_robot, providing research capacity, engineering expertise, infrastructure, and operational support.
 

diff --git a/docs/en/getting-started/installation.md b/docs/en/getting-started/installation.md
@@ -135,7 +135,7 @@ docker build -t sgr-agent-core:latest .
 
 ## Configuration
 
-After installation, you'll need to configure your API keys and settings. See the [Configuration Guide](../framework/configuration.md) for detailed instructions.
+After installation, you'll need to configure your API keys and settings. See the [Configuration Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md) for detailed instructions.
 
 ### Quick Configuration
 
@@ -162,5 +162,5 @@ export SGR__LLM__MODEL="gpt-4o"
 
 ## Next Steps
 
-* **[Quick Start Guide](../framework/first-steps.md)** — Get started with your first agent
-* **[Configuration Guide](../framework/configuration.md)** — Configure your agents
+* **[Quick Start Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/first-steps.md)** — Get started with your first agent
+* **[Configuration Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md)** — Configure your agents
diff --git a/docs/en/highlights.md b/docs/en/highlights.md
@@ -3,9 +3,9 @@
 This page answers one simple question: **What's cool about this framework?**
 Short cases and intuitive examples; for more complete and detailed docs go to:
 
-- [Main Concepts](framework/main-concepts.md)
-- [Configuration Guide](framework/configuration.md)
-- [Getting Started](getting-started/index.md)
+- [Main Concepts](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/main-concepts.md)
+- [Configuration Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md)
+- [Getting Started](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/getting-started/index.md)
 
 ---
 
@@ -69,7 +69,7 @@ The idea: you describe an agent (in code or a config file) and run it for a spec
 **-->RTFM**
 
 - install `sgr-agent-core`;
-- read the [configuration](framework/configuration.md) and [main concepts](framework/main-concepts.md) sections;
+- read the [configuration](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md) and [main concepts](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/main-concepts.md) sections;
 - describe agents via `config.yaml` and `agents.yaml` (or any other convenient form);
 - extend base entities in code to fit your agent's logic and tools;
 - call `execute()`.
@@ -105,7 +105,7 @@ Think of it like docker compose — **one YAML that describes what to do**:
 4. An API spins up — by default at `http://localhost:8010`, Swagger at `http://localhost:8010/docs`.
 5. Send requests using the OpenAI chat/completions protocol.
 
-More details — [API Server Quick Start](sgr-api/SGR-Quick-Start.md).
+More details — [API Server Quick Start](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/sgr-api/SGR-Quick-Start.md).
 
 ---
 
@@ -184,7 +184,7 @@ docker run -d \
 
 After startup open `http://localhost:3000` — agents from your config will appear in the models list.
 
-![Open WebUI with SGR agent](../assets/images/openwebui_example.png)
+![Open WebUI with SGR agent](openwebui_example.png)
 
 > **Important:** when connecting to Open WebUI you need to switch the streaming adapter. In `config.yaml` set:
 > ```yaml
@@ -233,7 +233,7 @@ agents:
       - "final_answer_tool"
 ```
 
-Full configuration schema and examples in [Configuration Guide](framework/configuration.md).
+Full configuration schema and examples in [Configuration Guide](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/en/framework/configuration.md).
 
 ---
 

diff --git a/docs/ru/framework/agents.md b/docs/ru/framework/agents.md
@@ -3,7 +3,7 @@
 В этом разделе погрузимся в техническую реализацию агента и разберём подробнее, как это можно адаптировать под свои нужды
 
 !!! Tip "Архитектура агентов"
-    Информация о доступных агентах и их различиях находится в [Основных концепциях](main-concepts.md#agent).
+    Информация о доступных агентах и их различиях находится в [Основных концепциях](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/ru/framework/main-concepts.md#agent).
 
 
 *Для детального понимания полной логики лучше ознакомиться с [исходным кодом](https://github.com/vamplabAI/sgr-agent-core/blob/main/sgr_agent_core/base_agent.py).*
@@ -137,7 +137,7 @@ class BaseAgent:
 Хранит все настройки агента: параметры LLM, поиска, выполнения, промпты и MCP конфигурацию.
 
 !!! Tip "Подробнее о конфигурации"
-    Полное описание системы конфигурации, иерархии настроек и примеры использования находятся в [руководстве по конфигурации](configuration.md).
+    Полное описание системы конфигурации, иерархии настроек и примеры использования находятся в [руководстве по конфигурации](Соцсети/_архив/AiConf/aiconf/sgr-agent-core/docs/ru/framework/configuration.md).
 
 #### Расширение конфигурации