ao-dev 0.0.3__tar.gz

ao_dev-0.0.3/PKG-INFO

@@ -0,0 +1,41 @@
+Metadata-Version: 2.4
+Name: ao-dev
+Version: 0.0.3
+Summary: A development tool for LLM call visualization and debugging
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: pyyaml
+Requires-Dist: dill
+Requires-Dist: debugpy
+Requires-Dist: psutil
+Requires-Dist: psycopg2-binary
+Requires-Dist: python-dotenv
+Requires-Dist: fastapi
+Requires-Dist: flatten_dict
+Requires-Dist: flatten_json
+Provides-Extra: dev
+Requires-Dist: black==25.1.0; extra == "dev"
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: pre-commit; extra == "dev"
+Requires-Dist: responses; extra == "dev"
+Requires-Dist: openai; extra == "dev"
+Requires-Dist: anthropic; extra == "dev"
+Requires-Dist: openai-agents; extra == "dev"
+Requires-Dist: google-genai; extra == "dev"
+Requires-Dist: hydra-core; extra == "dev"
+Requires-Dist: e2b_code_interpreter; extra == "dev"
+Requires-Dist: together; extra == "dev"
+Requires-Dist: langchain; extra == "dev"
+Requires-Dist: langchain-anthropic; extra == "dev"
+Requires-Dist: langchain-openai; extra == "dev"
+Provides-Extra: examples
+Requires-Dist: openai; extra == "examples"
+Requires-Dist: anthropic; extra == "examples"
+Requires-Dist: openai-agents; extra == "examples"
+Requires-Dist: google-genai; extra == "examples"
+Provides-Extra: docs
+Requires-Dist: mkdocs-material>=9.5; extra == "docs"
+Requires-Dist: mkdocs-minify-plugin; extra == "docs"
+Requires-Dist: mkdocstrings[python]>=0.24; extra == "docs"
+
+This tool allows you to visualize and edit your agent trajectories. Check out https://github.com/agent-ops-project/agent-dev for more details!

ao_dev-0.0.3/README.md

@@ -0,0 +1,87 @@
+# AO Agent Dev: What-if Questions over Agent Trajectories  
+
+AO Agent Dev is a developer tool for agent builders. It supports arbitrary Python progams (like your existing ones!) and will visualize your agent's LLM and MCP calls into a dataflow graph. You can then inspect, edit and rerun this graph and understand how to fix your agent.
+
+[![Quickstart video](docs/media/quickstart_screenshot.png)](https://youtu.be/woVctiSRJU0)
+
+
+## Quickstart
+
+`AO Agent Dev` integrates into VS Code or IDEs based on it, like Cursor. Simply download our [VS Code Extension](https://marketplace.visualstudio.com/items?itemName=agentops.ao-agent-dev) and install our pip package:
+
+```bash
+pip install ao-agent-dev
+```
+
+Then, give it a spin:
+
+#### 1. Create an Example Project
+
+Create a folder called `my-agent` and add a file called `openai_example.py` with the following content:
+
+```python
+from openai import OpenAI
+
+def main():
+    client = OpenAI()
+
+    response = client.responses.create(
+        model="gpt-4o-mini",
+        input="Output the number 42 and nothing else",
+        temperature=0
+    )
+    number = response.output_text
+
+    prompt_add_1 = f"Add 1 to {number} and just output the result."
+    prompt_add_2 = f"Add 2 to {number} and just output the result."
+
+    response1 = client.responses.create(model="gpt-4o-mini", input=prompt_add_1, temperature=0)
+    response2 = client.responses.create(model="gpt-4o-mini", input=prompt_add_2, temperature=0)
+
+    sum_prompt = f"Add these two numbers together and just output the result: {response1.output_text} + {response2.output_text}"
+    final_sum = client.responses.create(model="gpt-4o-mini", input=sum_prompt, temperature=0)
+
+    print(f"Final sum: {final_sum.output_text}")
+
+if __name__ == "__main__":
+    main()
+```
+
+Run the script to verify it works:
+
+```bash
+cd my-agent
+python openai_example.py
+```
+
+#### 2. Configure AO
+
+Run `ao-config` and set the project root to your `my-agent` folder:
+
+```bash
+ao-config
+```
+
+#### 3. Run an example
+
+Open your `my-agent` folder in VS Code, then run the example with AO in the terminal:
+
+```bash
+ao-record openai_example.py
+```
+
+The VS Code extension will display the dataflow graph showing how data flows between the LLM calls.
+
+## Documentation
+
+For complete documentation, installation guides, and tutorials, visit our **[Documentation Site](https://ao-agent-ops.github.io/ao-dev/)**.
+
+## Building from source and developing
+
+See the [Installation Guide](https://ao-agent-ops.github.io/ao-dev/getting-started/installation/) for development setup and the [Developer Guide](https://ao-agent-ops.github.io/ao-dev/developer-guide/architecture/) for architecture details.
+
+## Community
+
+- [Join our Discord](https://discord.gg/fjsNSa6TAh)
+- [GitHub Issues](https://github.com/agent-ops-project/ao-dev/issues)
+- We're just getting started on this tool and are eager to hear your feedback and resolve issues you ran into! We hope you enjoy it as much as we do.

ao_dev-0.0.3/ao_dev.egg-info/PKG-INFO

@@ -0,0 +1,41 @@
+Metadata-Version: 2.4
+Name: ao-dev
+Version: 0.0.3
+Summary: A development tool for LLM call visualization and debugging
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+Requires-Dist: pyyaml
+Requires-Dist: dill
+Requires-Dist: debugpy
+Requires-Dist: psutil
+Requires-Dist: psycopg2-binary
+Requires-Dist: python-dotenv
+Requires-Dist: fastapi
+Requires-Dist: flatten_dict
+Requires-Dist: flatten_json
+Provides-Extra: dev
+Requires-Dist: black==25.1.0; extra == "dev"
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: pre-commit; extra == "dev"
+Requires-Dist: responses; extra == "dev"
+Requires-Dist: openai; extra == "dev"
+Requires-Dist: anthropic; extra == "dev"
+Requires-Dist: openai-agents; extra == "dev"
+Requires-Dist: google-genai; extra == "dev"
+Requires-Dist: hydra-core; extra == "dev"
+Requires-Dist: e2b_code_interpreter; extra == "dev"
+Requires-Dist: together; extra == "dev"
+Requires-Dist: langchain; extra == "dev"
+Requires-Dist: langchain-anthropic; extra == "dev"
+Requires-Dist: langchain-openai; extra == "dev"
+Provides-Extra: examples
+Requires-Dist: openai; extra == "examples"
+Requires-Dist: anthropic; extra == "examples"
+Requires-Dist: openai-agents; extra == "examples"
+Requires-Dist: google-genai; extra == "examples"
+Provides-Extra: docs
+Requires-Dist: mkdocs-material>=9.5; extra == "docs"
+Requires-Dist: mkdocs-minify-plugin; extra == "docs"
+Requires-Dist: mkdocstrings[python]>=0.24; extra == "docs"
+
+This tool allows you to visualize and edit your agent trajectories. Check out https://github.com/agent-ops-project/agent-dev for more details!

ao_dev-0.0.3/ao_dev.egg-info/SOURCES.txt

@@ -0,0 +1,50 @@
+README.md
+pyproject.toml
+ao_dev.egg-info/PKG-INFO
+ao_dev.egg-info/SOURCES.txt
+ao_dev.egg-info/dependency_links.txt
+ao_dev.egg-info/entry_points.txt
+ao_dev.egg-info/requires.txt
+ao_dev.egg-info/top_level.txt
+docs/PKG_README.md
+src/__init__.py
+src/cli/__init__.py
+src/cli/ao_config.py
+src/cli/ao_record.py
+src/cli/ao_server.py
+src/common/__init__.py
+src/common/config.py
+src/common/constants.py
+src/common/logger.py
+src/common/utils.py
+src/runner/__init__.py
+src/runner/agent_runner.py
+src/runner/ast_rewrite_hook.py
+src/runner/context_manager.py
+src/runner/taint_dict.py
+src/runner/monkey_patching/__init__.py
+src/runner/monkey_patching/api_parser.py
+src/runner/monkey_patching/apply_monkey_patches.py
+src/runner/monkey_patching/patching_utils.py
+src/runner/monkey_patching/api_parsers/__init__.py
+src/runner/monkey_patching/api_parsers/genai_api_parser.py
+src/runner/monkey_patching/api_parsers/httpx_api_parser.py
+src/runner/monkey_patching/api_parsers/mcp_api_parser.py
+src/runner/monkey_patching/api_parsers/requests_api_parser.py
+src/runner/monkey_patching/patches/__init__.py
+src/runner/monkey_patching/patches/file_patches.py
+src/runner/monkey_patching/patches/genai_patch.py
+src/runner/monkey_patching/patches/httpx_patch.py
+src/runner/monkey_patching/patches/mcp_patches.py
+src/runner/monkey_patching/patches/randomness_patch.py
+src/runner/monkey_patching/patches/requests_patch.py
+src/server/__init__.py
+src/server/ast_helpers.py
+src/server/ast_transformer.py
+src/server/auth_app.py
+src/server/database_manager.py
+src/server/file_watcher.py
+src/server/main_server.py
+src/server/database_backends/__init__.py
+src/server/database_backends/postgres.py
+src/server/database_backends/sqlite.py

ao_dev-0.0.3/ao_dev.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

ao_dev-0.0.3/ao_dev.egg-info/entry_points.txt

@@ -0,0 +1,6 @@
+[console_scripts]
+ao-config = ao.cli.ao_config:main
+ao-record = ao.cli.ao_record:main
+ao-server = ao.cli.ao_server:main
+record = ao.cli.ao_record:main
+server = ao.cli.ao_server:main

ao_dev-0.0.3/ao_dev.egg-info/requires.txt

@@ -0,0 +1,36 @@
+pyyaml
+dill
+debugpy
+psutil
+psycopg2-binary
+python-dotenv
+fastapi
+flatten_dict
+flatten_json
+
+[dev]
+black==25.1.0
+pytest
+pre-commit
+responses
+openai
+anthropic
+openai-agents
+google-genai
+hydra-core
+e2b_code_interpreter
+together
+langchain
+langchain-anthropic
+langchain-openai
+
+[docs]
+mkdocs-material>=9.5
+mkdocs-minify-plugin
+mkdocstrings[python]>=0.24
+
+[examples]
+openai
+anthropic
+openai-agents
+google-genai

ao_dev-0.0.3/ao_dev.egg-info/top_level.txt

@@ -0,0 +1 @@
+ao

ao_dev-0.0.3/docs/PKG_README.md

@@ -0,0 +1 @@
+This tool allows you to visualize and edit your agent trajectories. Check out https://github.com/agent-ops-project/agent-dev for more details!

ao_dev-0.0.3/pyproject.toml

@@ -0,0 +1,83 @@
+# pyproject.toml
+
+[build-system]
+requires = ["setuptools"]
+build-backend = "setuptools.build_meta"
+
+[tool.setuptools]
+# Map ao package to src directory with subpackages
+package-dir = {"ao" = "src"}
+packages = [
+    "ao",
+    "ao.cli",
+    "ao.common",
+    "ao.runner",
+    "ao.runner.monkey_patching",
+    "ao.runner.monkey_patching.api_parsers",
+    "ao.runner.monkey_patching.patches",
+    "ao.server",
+    "ao.server.database_backends"
+]
+
+[project.scripts]
+# Console commands for the ao package
+ao-record = "ao.cli.ao_record:main"
+ao-config = "ao.cli.ao_config:main"
+ao-server = "ao.cli.ao_server:main"
+
+# Aliases for quick access
+record = "ao.cli.ao_record:main"
+server = "ao.cli.ao_server:main"
+
+[tool.black]
+line-length = 100
+
+[project]
+name = "ao-dev"
+version = "0.0.3"
+description = "A development tool for LLM call visualization and debugging"
+readme = "docs/PKG_README.md"
+requires-python = ">=3.10"
+
+dependencies = [
+    "pyyaml",
+    "dill",
+    "debugpy",
+    "psutil",
+    "psycopg2-binary",
+    "python-dotenv",
+    "fastapi",
+    "flatten_dict",
+    "flatten_json"
+]
+
+[project.optional-dependencies]
+dev = [
+    "black==25.1.0",
+    "pytest",
+    "pre-commit",
+    "responses",
+    "openai",
+    "anthropic",
+    "openai-agents",
+    "google-genai",
+    "hydra-core",
+    "e2b_code_interpreter",
+    "together",
+    "langchain",
+    "langchain-anthropic",
+    "langchain-openai"
+]
+
+examples = [
+    "openai",
+    "anthropic",
+    "openai-agents",
+    "google-genai"
+]
+
+docs = [
+    "mkdocs-material>=9.5",
+    "mkdocs-minify-plugin",
+    "mkdocstrings[python]>=0.24",
+]

ao_dev-0.0.3/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

ao_dev-0.0.3/src/__init__.py

@@ -0,0 +1,3 @@
+from ao.runner.context_manager import ao_launch as launch, log
+
+__all__ = ["launch", "log"]

ao_dev-0.0.3/src/cli/ao_config.py

@@ -0,0 +1,71 @@
+# Register taint functions in builtins BEFORE any other imports
+import builtins
+from ao.server.ast_helpers import (
+    taint_fstring_join,
+    taint_format_string,
+    taint_percent_format,
+    exec_func,
+)
+
+builtins.taint_fstring_join = taint_fstring_join
+builtins.taint_format_string = taint_format_string
+builtins.taint_percent_format = taint_percent_format
+builtins.exec_func = exec_func
+
+import argparse
+import os
+from ao.common.config import (
+    Config,
+    _ask_field,
+    _convert_to_valid_path,
+)
+from ao.common.constants import AO_CONFIG, AO_PROJECT_ROOT
+
+
+def get_user_input() -> Config:
+    project_root = _ask_field(
+        f"What is the root directory of your project? [{AO_PROJECT_ROOT}]\n> ",
+        _convert_to_valid_path,
+        default=AO_PROJECT_ROOT,
+        error_message="Please enter a valid path to a directory.",
+    )
+
+    database_url = _ask_field(
+        "Database URL (leave empty for SQLite)\n> ",
+        str,
+        default=os.environ.get("DATABASE_URL"),
+        error_message="Please enter a valid database URL or leave empty.",
+    )
+
+    config = Config(
+        project_root=project_root,
+        database_url=database_url,
+    )
+    return config
+
+
+def config_command():
+    config = get_user_input()
+    config_file = AO_CONFIG
+    config.to_yaml_file(config_file)
+
+
+def config_command_parser():
+    description = (
+        "Run `ao config` before you debug your agents. This "
+        "will prompt some configurations that you can choose. "
+        "These will get saved in a default path or in --config_path "
+        "which you can pass: `ao config --config_path some/path/config.yaml"
+    )
+    parser = argparse.ArgumentParser("Config", usage="ao-config", description=description)
+    return parser
+
+
+def main():
+    parser = config_command_parser()
+    parser.parse_args()
+    config_command()
+
+
+if __name__ == "__main__":
+    main()

ao_dev-0.0.3/src/cli/ao_record.py

@@ -0,0 +1,92 @@
+import os
+import time
+import yaml
+from argparse import ArgumentParser, REMAINDER
+from ao.common.constants import AO_CONFIG
+from ao.runner.agent_runner import AgentRunner
+
+
+def launch_command_parser():
+    parser = ArgumentParser(
+        usage="ao-record <script.py> [<args>]",
+        description="Launch a python script with the AO under the hood.",
+        allow_abbrev=False,
+    )
+
+    parser.add_argument(
+        "--config-file",
+        default=None,
+        help="The config file to use for the default values in the launching script.",
+    )
+
+    parser.add_argument(
+        "--run-name",
+        default=None,
+        help="Name that will be used in the experiment list. If not set, Run X where X is the index of the current run will be used.",
+    )
+
+    parser.add_argument(
+        "-m",
+        "--module",
+        action="store_true",
+        help="Change each process to interpret the launch script as a Python module, executing with the same behavior as 'python -m'.",
+    )
+
+    parser.add_argument(
+        "script_path",
+        type=str,
+        help="The full path to the script to be executed, followed by all the remaining arguments.",
+    )
+
+    parser.add_argument(
+        "script_args", nargs=REMAINDER, help="Arguments of the script to be executed."
+    )
+    return parser
+
+
+def _validate_launch_command(args):
+    default_dict = {}
+    config_file = None
+
+    # check if the location of the config file is set
+    if args.config_file is not None and os.path.isfile(args.config_file):
+        config_file = args.config_file
+    # if not, check in the AOC_CONFIG path
+    elif os.path.isfile(AO_CONFIG):
+        config_file = AO_CONFIG
+    # if also not, it stays empty and we rely on the (default) args
+
+    if config_file is not None:
+        with open(config_file, encoding="utf-8") as f:
+            default_dict = yaml.safe_load(f)
+
+    # the arguments overwrite what is written in the config file
+    # that's why they come second.
+    args.__dict__ = {**default_dict, **args.__dict__}
+    # we don't need the config_file anymore and we don't
+    # want to confuse people with it still in the args:
+    del args.config_file
+
+    return args
+
+
+def launch_command(args):
+    args = _validate_launch_command(args)
+
+    agent_runner = AgentRunner(
+        script_path=args.script_path,
+        script_args=args.script_args,
+        is_module_execution=args.module,
+        run_name=args.run_name,
+    )
+    agent_runner.run()
+
+
+def main():
+    parser = launch_command_parser()
+    args = parser.parse_args()
+    launch_command(args)
+
+
+if __name__ == "__main__":
+    main()