syna-agent 0.9.0__tar.gz

syna_agent-0.9.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Julynx
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

syna_agent-0.9.0/PKG-INFO

@@ -0,0 +1,132 @@
+Metadata-Version: 2.4
+Name: syna-agent
+Version: 0.9.0
+Summary: Terminal and web AI agent that answers requests by driving an LLM in a tool-call loop inside a sandboxed container pod.
+Keywords: ai,agent,llm,tools,sandbox,openrouter,assistant
+Author: Julynx
+Author-email: Julynx <julicabria@outlook.com>
+License-Expression: MIT
+License-File: LICENSE
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Console
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Developers
+Classifier: Intended Audience :: End Users/Desktop
+Classifier: Operating System :: Microsoft :: Windows
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Operating System :: MacOS
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Dist: openrouter>=1.1.137
+Requires-Dist: python-dotenv>=1.2.3
+Requires-Dist: httpx>=0.27
+Requires-Dist: bollard>=0.1.3
+Requires-Dist: playwright==1.62.0
+Requires-Dist: drissionpage>=4.1.1.4
+Requires-Dist: markdownify>=1.2.3
+Requires-Dist: pillow>=11.0.0
+Requires-Dist: fastapi>=0.115.0
+Requires-Dist: uvicorn[standard]>=0.30.0
+Requires-Dist: python-multipart>=0.0.9
+Requires-Dist: camoufox[geoip]>=0.5.8
+Requires-Python: >=3.12
+Project-URL: Homepage, https://julynx.github.io/syna/
+Project-URL: Repository, https://github.com/Julynx/syna
+Project-URL: Issues, https://github.com/Julynx/syna/issues
+Description-Content-Type: text/markdown
+
+<p align="center">
+  <img
+    src="https://raw.githubusercontent.com/Julynx/syna/main/website/assets/syna-logo-pill.png"
+    width="260px"
+  />
+</p>
+
+<h3 align="center">Syna</h3>
+<p align="center">
+  <i>The AI agent with a pod.</i>
+  <br />
+  <a href="https://julynx.github.io/syna/" target="_blank">julynx.github.io/syna</a>
+</p>
+
+<hr />
+
+<p align="center">
+  <h3 align="center"><code>uv tool install syna-agent</code></h3>
+</p>
+
+![Syna chat interface](website/assets/screenshot.png)
+
+![Syna voice mode with the ambient visualizer](website/assets/screenshot2.png)
+
+## Features
+
+- **A sandboxed pod.** Every command runs inside an isolated container, with a full Debian shell, root access, and your host machine out of reach.
+- **A complete toolset.** Web search and browser automation (both local, no API key), image understanding, image generation, speech-to-text, text-to-speech, and precise line-number, diff-style file editing.
+- **Built for long, complex tasks.** Every tool output is saved to a file, context is managed automatically (elision and compaction, with a high cache hit rate), and deliverables keep a verifiable chain of custody from source to artifact.
+- **Honest reporting.** Syna is trained to say exactly what happened and what it can't verify; it resists prompt injection and false-flag attacks baked into web content, and it scores at the top of honesty benchmarks.
+- **It asks when it should.** Multiple-choice or open-ended questions, mid-task, whenever a decision isn't Syna's to make.
+- **Voice mode.** Talk to it, and it reads its answers out loud.
+- **A beautiful UI.** Frosted glass, sound effects, and a Y2K-video-player-style background visualizer that reacts to the agent (thinking, calling tools, speaking).
+- **Any provider.** Works with OpenRouter (recommended, most tested), OpenAI, Google Gemini, local Ollama, or any OpenAI-compatible endpoint.
+
+## Getting started
+
+1. Install:
+
+   ```bash
+   uv tool install syna-agent
+   ```
+
+2. Run the setup wizard and follow the steps (dependencies, provider, API key, sanity check):
+
+   ```bash
+   syna-setup
+   ```
+
+3. Start Syna (the terminal version, or the web UI at `http://127.0.0.1:8000`):
+
+   ```bash
+   syna
+   syna-server
+   ```
+
+The wizard writes a `.env` file in `~/.syna` for you. To configure manually, create `~/.syna/.env` with:
+
+```sh
+OPENROUTER_API_KEY=your-api-key
+MODEL=z-ai/glm-5.3-flash        # Or any one of your preference
+BASE_URL=https://openrouter.ai/api/v1   # Optional: any OpenAI-compatible endpoint
+THINKING_EFFORT=medium          # Can be either of none|minimal|low|medium|high|xhigh|max
+```
+
+Voice mode is enabled by adding speech models to the same `.env` file:
+
+```sh
+STT_MODEL=google/chirp-3                    # Transcribes microphone recordings
+TTS_MODEL=google/gemini-3.8-flash-lite-tts  # Reads the answers out loud
+TTS_VOICE=Kore                              # Optional: voice of the TTS model
+SPEECH_LANGUAGE=en                          # Optional: ISO-639-1 code, English by default
+```
+
+Image generation (the model's `text_to_image` tool) is enabled the same way:
+
+```sh
+IMAGE_MODEL=google/gemini-3.1-flash-image
+```
+
+Requirements: Python 3.12+ and a container engine (Podman or Docker; the setup wizard checks and installs one for you).
+
+## Privacy
+
+Syna does not collect your personal information or send your messages anywhere except to the provider you specified. Their Terms of Service and Privacy Policy apply.
+
+## Links
+
+- [Website](https://julynx.github.io/syna/)
+- [PyPI](https://pypi.org/project/syna-agent/)
+- [Technical description](docs/technical_description.md)
+- [Report an issue](https://github.com/Julynx/syna/issues)

syna_agent-0.9.0/README.md

@@ -0,0 +1,92 @@
+<p align="center">
+  <img
+    src="https://raw.githubusercontent.com/Julynx/syna/main/website/assets/syna-logo-pill.png"
+    width="260px"
+  />
+</p>
+
+<h3 align="center">Syna</h3>
+<p align="center">
+  <i>The AI agent with a pod.</i>
+  <br />
+  <a href="https://julynx.github.io/syna/" target="_blank">julynx.github.io/syna</a>
+</p>
+
+<hr />
+
+<p align="center">
+  <h3 align="center"><code>uv tool install syna-agent</code></h3>
+</p>
+
+![Syna chat interface](website/assets/screenshot.png)
+
+![Syna voice mode with the ambient visualizer](website/assets/screenshot2.png)
+
+## Features
+
+- **A sandboxed pod.** Every command runs inside an isolated container, with a full Debian shell, root access, and your host machine out of reach.
+- **A complete toolset.** Web search and browser automation (both local, no API key), image understanding, image generation, speech-to-text, text-to-speech, and precise line-number, diff-style file editing.
+- **Built for long, complex tasks.** Every tool output is saved to a file, context is managed automatically (elision and compaction, with a high cache hit rate), and deliverables keep a verifiable chain of custody from source to artifact.
+- **Honest reporting.** Syna is trained to say exactly what happened and what it can't verify; it resists prompt injection and false-flag attacks baked into web content, and it scores at the top of honesty benchmarks.
+- **It asks when it should.** Multiple-choice or open-ended questions, mid-task, whenever a decision isn't Syna's to make.
+- **Voice mode.** Talk to it, and it reads its answers out loud.
+- **A beautiful UI.** Frosted glass, sound effects, and a Y2K-video-player-style background visualizer that reacts to the agent (thinking, calling tools, speaking).
+- **Any provider.** Works with OpenRouter (recommended, most tested), OpenAI, Google Gemini, local Ollama, or any OpenAI-compatible endpoint.
+
+## Getting started
+
+1. Install:
+
+   ```bash
+   uv tool install syna-agent
+   ```
+
+2. Run the setup wizard and follow the steps (dependencies, provider, API key, sanity check):
+
+   ```bash
+   syna-setup
+   ```
+
+3. Start Syna (the terminal version, or the web UI at `http://127.0.0.1:8000`):
+
+   ```bash
+   syna
+   syna-server
+   ```
+
+The wizard writes a `.env` file in `~/.syna` for you. To configure manually, create `~/.syna/.env` with:
+
+```sh
+OPENROUTER_API_KEY=your-api-key
+MODEL=z-ai/glm-5.3-flash        # Or any one of your preference
+BASE_URL=https://openrouter.ai/api/v1   # Optional: any OpenAI-compatible endpoint
+THINKING_EFFORT=medium          # Can be either of none|minimal|low|medium|high|xhigh|max
+```
+
+Voice mode is enabled by adding speech models to the same `.env` file:
+
+```sh
+STT_MODEL=google/chirp-3                    # Transcribes microphone recordings
+TTS_MODEL=google/gemini-3.8-flash-lite-tts  # Reads the answers out loud
+TTS_VOICE=Kore                              # Optional: voice of the TTS model
+SPEECH_LANGUAGE=en                          # Optional: ISO-639-1 code, English by default
+```
+
+Image generation (the model's `text_to_image` tool) is enabled the same way:
+
+```sh
+IMAGE_MODEL=google/gemini-3.1-flash-image
+```
+
+Requirements: Python 3.12+ and a container engine (Podman or Docker; the setup wizard checks and installs one for you).
+
+## Privacy
+
+Syna does not collect your personal information or send your messages anywhere except to the provider you specified. Their Terms of Service and Privacy Policy apply.
+
+## Links
+
+- [Website](https://julynx.github.io/syna/)
+- [PyPI](https://pypi.org/project/syna-agent/)
+- [Technical description](docs/technical_description.md)
+- [Report an issue](https://github.com/Julynx/syna/issues)

syna_agent-0.9.0/pyproject.toml

@@ -0,0 +1,67 @@
+[project]
+name = "syna-agent"
+version = "0.9.0"
+description = "Terminal and web AI agent that answers requests by driving an LLM in a tool-call loop inside a sandboxed container pod."
+readme = "README.md"
+license = "MIT"
+license-files = ["LICENSE"]
+requires-python = ">=3.12"
+keywords = [
+    "ai",
+    "agent",
+    "llm",
+    "tools",
+    "sandbox",
+    "openrouter",
+    "assistant",
+]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Environment :: Console",
+    "Environment :: Web Environment",
+    "Intended Audience :: Developers",
+    "Intended Audience :: End Users/Desktop",
+    "Operating System :: Microsoft :: Windows",
+    "Operating System :: POSIX :: Linux",
+    "Operating System :: MacOS",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+dependencies = [
+    "openrouter>=1.1.137",
+    "python-dotenv>=1.2.3",
+    "httpx>=0.27",
+    "bollard>=0.1.3",
+    "playwright==1.62.0",
+    "drissionpage>=4.1.1.4",
+    "markdownify>=1.2.3",
+    "pillow>=11.0.0",
+    "fastapi>=0.115.0",
+    "uvicorn[standard]>=0.30.0",
+    "python-multipart>=0.0.9",
+    "camoufox[geoip]>=0.5.8",
+]
+
+[[project.authors]]
+name = "Julynx"
+email = "julicabria@outlook.com"
+
+[project.scripts]
+syna = "syna:main"
+syna-server = "syna.server:main"
+syna-setup = "syna.setup:main"
+
+[project.urls]
+Homepage = "https://julynx.github.io/syna/"
+Repository = "https://github.com/Julynx/syna"
+Issues = "https://github.com/Julynx/syna/issues"
+
+[build-system]
+requires = ["uv_build>=0.12.3,<0.13.0"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-name = "syna"

syna_agent-0.9.0/pyproject.toml.orig

@@ -0,0 +1,58 @@
+[project]
+name = "syna-agent"
+version = "0.9.0"
+description = "Terminal and web AI agent that answers requests by driving an LLM in a tool-call loop inside a sandboxed container pod."
+readme = "README.md"
+license = "MIT"
+license-files = ["LICENSE"]
+authors = [
+    { name = "Julynx", email = "julicabria@outlook.com" }
+]
+requires-python = ">=3.12"
+keywords = ["ai", "agent", "llm", "tools", "sandbox", "openrouter", "assistant"]
+classifiers = [
+    "Development Status :: 4 - Beta",
+    "Environment :: Console",
+    "Environment :: Web Environment",
+    "Intended Audience :: Developers",
+    "Intended Audience :: End Users/Desktop",
+    "Operating System :: Microsoft :: Windows",
+    "Operating System :: POSIX :: Linux",
+    "Operating System :: MacOS",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.12",
+    "Programming Language :: Python :: 3.13",
+    "Programming Language :: Python :: 3.14",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+dependencies = [
+    "openrouter>=1.1.137",
+    "python-dotenv>=1.2.3",
+    "httpx>=0.27",
+    "bollard>=0.1.3",
+    "playwright==1.62.0",
+    "drissionpage>=4.1.1.4",
+    "markdownify>=1.2.3",
+    "pillow>=11.0.0",
+    "fastapi>=0.115.0",
+    "uvicorn[standard]>=0.30.0",
+    "python-multipart>=0.0.9",
+    "camoufox[geoip]>=0.5.8",
+]
+
+[project.scripts]
+syna = "syna:main"
+syna-server = "syna.server:main"
+syna-setup = "syna.setup:main"
+
+[project.urls]
+Homepage = "https://julynx.github.io/syna/"
+Repository = "https://github.com/Julynx/syna"
+Issues = "https://github.com/Julynx/syna/issues"
+
+[build-system]
+requires = ["uv_build>=0.12.3,<0.13.0"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-name = "syna"

syna_agent-0.9.0/src/syna/__init__.py

@@ -0,0 +1,22 @@
+"""Application entrypoint for Syna AI agent."""
+
+import sys
+
+from .ask import ask_loop
+from .state import close_docker_client
+
+
+def main() -> None:
+    """Start the interactive agent loop."""
+    try:
+        ask_loop()
+    except KeyboardInterrupt:
+        print("\n  + (Received CTRL+C)")
+        sys.exit(0)
+    finally:
+        try:
+            print("  + (Cleaning up container and shutting down, please wait...)")
+            close_docker_client()
+            print("  + (Cleanup finished. Goodbye!)\n")
+        except KeyboardInterrupt:
+            pass

syna_agent-0.9.0/src/syna/ask.py

@@ -0,0 +1,162 @@
+"""CLI interaction loop built on top of the shared AgentSession."""
+
+import sys
+
+from .ask_utils import show_welcome
+from .commands import parse_and_execute_command
+from .interaction import InteractionCancelled, register_responder
+from .interaction import unregister_responder
+from .session import AgentSession
+from .state import reset_shared_container
+
+if hasattr(sys.stdout, "reconfigure"):
+    sys.stdout.reconfigure(encoding="utf-8", errors="replace")
+if hasattr(sys.stderr, "reconfigure"):
+    sys.stderr.reconfigure(encoding="utf-8", errors="replace")
+
+NEW_CHAT_COMMAND = "/newchat"
+
+
+def _print_status_line(text: str):
+    """Overwrite the progress line so repeated statuses don't scroll."""
+    print(f"  + ({text})", end="\r", flush=True)
+
+
+def _print_tool_call(event):
+    """Render one tool_call event: started, finished, failed, or skipped."""
+    status = event.get("status")
+    signature = str(event.get("signature", "")).strip()[:64].replace("\n", " ")
+    if status == "started":
+        _print_status_line(f"Running {signature}...")
+        return
+    if status == "skipped":
+        print(f"  + │Skipped '{signature}...' (an earlier call in the batch failed)")
+        _print_status_line("Thinking...")
+        return
+    if status != "finished":
+        return
+    shared = event.get("structured_output")
+    if isinstance(shared, dict) and "path" in shared:
+        print(f"  + │Shared '{shared.get('name', '')}'")
+        print(f"    └─ Save it locally with: /get-from {shared['path']}")
+        _print_status_line("Thinking...")
+        return
+    output = str(event.get("output_preview", "")).strip()[:64]
+    if event.get("failed"):
+        print(f"  + ✗Failed '{signature}...'")
+    else:
+        print(f"  + │Used tool '{signature}...'")
+    print(f"    └─ {output}...")
+    _print_status_line("Thinking...")
+
+
+def _print_event(event):
+    """Render session events with the classic CLI look and feel."""
+    kind = event["type"]
+    if kind == "status":
+        _print_status_line(event["text"])
+    elif kind == "tool_call":
+        _print_tool_call(event)
+    elif kind == "model_message":
+        print(f"\n[Syna]: {event['content']}\n")
+    elif kind == "compaction":
+        print(f"\n----- Compaction -----\n{event['summary']}\n")
+    elif kind == "error":
+        print(f"\n  ! ({event['text']})")
+
+
+def _terminal_responder(session: AgentSession):
+    """Build an ask_user responder that reads answers from the terminal.
+
+    A number picks one of the shown choices; any other text is delivered
+    as a custom answer. Ctrl+C (or a closed stdin) cancels the pending
+    question and stops the task instead of killing the application.
+    """
+
+    def responder(question_text: str, choices: list[str]) -> str:
+        print(f"\n  ? {question_text}")
+        if choices:
+            for number, choice in enumerate(choices, start=1):
+                print(f"    {number}. {choice}")
+            print("    Answer with a number, or type your own response.")
+        _print_status_line("Waiting for your answer...")
+        try:
+            while True:
+                answer = input("  > ").strip()
+                if not answer:
+                    continue
+                if choices and answer.isdigit() and 1 <= int(answer) <= len(choices):
+                    return choices[int(answer) - 1]
+                return answer
+        except (KeyboardInterrupt, EOFError):
+            session.cancel()
+            raise InteractionCancelled(
+                "The user stopped the task before answering."
+            ) from None
+
+    return responder
+
+
+def _handle_slash_command(query: str, session: AgentSession):
+    """Execute a slash command and echo its result or error."""
+    try:
+        result = parse_and_execute_command(query, session=session)
+        print(f"  + ({result})")
+    except ValueError as exc:
+        print(f"  ! ({exc})")
+
+
+def _create_session() -> AgentSession:
+    """Create an agent session that renders its events on the terminal."""
+    return AgentSession(on_event=_print_event, persist_outputs=True)
+
+
+def _reset_session(session: AgentSession) -> AgentSession:
+    """Drop the conversation and destroy the shared pod (/newchat)."""
+    print("  + (Cleaning up the pod, please wait...)")
+    try:
+        session.close()
+    except Exception:
+        pass
+    try:
+        reset_shared_container()
+    except Exception:
+        print("  ! (Could not clean up the old pod; starting a new session anyway.)")
+    print("  + (Started a fresh chat.)")
+    return _create_session()
+
+
+def ask_loop():
+    """Execute the interactive conversation and tool execution loop."""
+    show_welcome()
+    session = _create_session()
+    register_responder(_terminal_responder(session))
+    try:
+        while True:
+            try:
+                query = input("(Ask Syna): ")
+            except EOFError:
+                break
+            if not query.strip():
+                continue
+            if query.strip() == NEW_CHAT_COMMAND:
+                session = _reset_session(session)
+                register_responder(_terminal_responder(session))
+                continue
+            if query.strip().startswith("/"):
+                _handle_slash_command(query, session)
+                continue
+            session.clear_cancel()
+            try:
+                session.run_turn(query)
+            except KeyboardInterrupt:
+                session.cancel()
+                print("\n  + (Task stopped. Press CTRL+C again to quit.)")
+            except InteractionCancelled:
+                print("\n  + (Task stopped.)")
+    finally:
+        unregister_responder()
+        try:
+            session.close()
+        except Exception:
+            pass