AI agents suck. Weโre fixing that.
๐ฆ Twitter โข ๐ข Discord โข ๐๏ธ AgentOps โข ๐ Documentation
AgentOps helps developers build, evaluate, and monitor AI agents. Tools to build agents from prototype to production.
๐ Replay Analytics and Debugging | Step-by-step agent execution graphs |
๐ธ LLM Cost Management | Track spend with LLM foundation model providers |
๐งช Agent Benchmarking | Test your agents against 1,000+ evals |
๐ Compliance and Security | Detect common prompt injection and data exfiltration exploits |
๐ค Framework Integrations | Native Integrations with CrewAI, AutoGen, & LangChain |
pip install agentops
Initialize the AgentOps client and automatically get analytics on every LLM call.
import agentops
# Beginning of program's code (i.e. main.py, __init__.py)
agentops.init(<INSERT YOUR API KEY HERE>)
...
# (optional: record specific functions)
@agentops.record_function('sample function being record')
def sample_function(...):
...
# End of program
agentops.end_session('Success')
# Woohoo You're done ๐
All your sessions are available on the AgentOps dashboard. Refer to our API documentation for detailed instructions.
Build Crew agents with observability with only 2 lines of code. Simply set an AGENTOPS_API_KEY
in your environment, and your crews will get automatic monitoring on the AgentOps dashboard.
AgentOps is integrated with CrewAI on a pre-release fork. Install crew with
pip install git+https://github.com/AgentOps-AI/crewAI.git@main
With only two lines of code, add full observability and monitoring to Autogen agents. Set an AGENTOPS_API_KEY
in your environment and call agentops.init()
AgentOps works seamlessly with applications built using Langchain. To use the handler, install Langchain as an optional dependency:
Installation
pip install agentops[langchain]
To use the handler, import and set
import os
from langchain.chat_models import ChatOpenAI
from langchain.agents import initialize_agent, AgentType
from agentops.langchain_callback_handler import LangchainCallbackHandler
AGENTOPS_API_KEY = os.environ['AGENTOPS_API_KEY']
handler = LangchainCallbackHandler(api_key=AGENTOPS_API_KEY, tags=['Langchain Example'])
llm = ChatOpenAI(openai_api_key=OPENAI_API_KEY,
callbacks=[handler],
model='gpt-3.5-turbo')
agent = initialize_agent(tools,
llm,
agent=AgentType.CHAT_ZERO_SHOT_REACT_DESCRIPTION,
verbose=True,
callbacks=[handler], # You must pass in a callback handler to record your agent
handle_parsing_errors=True)
Check out the Langchain Examples Notebook for more details including Async handlers.
First class support for Cohere(>=5.4.0). This is a living integration, should you need any added functionality please message us on Discord!
Installation
pip install cohere
import cohere
import agentops
# Beginning of program's code (i.e. main.py, __init__.py)
agentops.init(<INSERT YOUR API KEY HERE>)
co = cohere.Client()
chat = co.chat(
message="Is it pronounced ceaux-hear or co-hehray?"
)
print(chat)
agentops.end_session('Success')
import cohere
import agentops
# Beginning of program's code (i.e. main.py, __init__.py)
agentops.init(<INSERT YOUR API KEY HERE>)
co = cohere.Client()
stream = co.chat_stream(
message="Write me a haiku about the synergies between Cohere and AgentOps"
)
for event in stream:
if event.event_type == "text-generation":
print(event.text, end='')
agentops.end_session('Success')
(Coming Soon)
(coming soon!)
(coming soon!)
Platform | Dashboard | Evals |
---|---|---|
โ Python SDK | โ Multi-session and Cross-session metrics | โ Custom eval metrics |
๐ง Evaluation builder API | โ Custom event tag tracking | ๐ Agent scorecards |
โ Javascript/Typescript SDK | โ Session replays | ๐ Evaluation playground + leaderboard |
Performance testing | Environments | LLM Testing | Reasoning and execution testing |
---|---|---|---|
โ Event latency analysis | ๐ Non-stationary environment testing | ๐ LLM non-deterministic function detection | ๐ง Infinite loops and recursive thought detection |
โ Agent workflow execution pricing | ๐ Multi-modal environments | ๐ง Token limit overflow flags | ๐ Faulty reasoning detection |
๐ง Success validators (external) | ๐ Execution containers | ๐ Context limit overflow flags | ๐ Generative code validators |
๐ Agent controllers/skill tests | โ Honeypot and prompt injection detection (PromptArmor) | ๐ API bill tracking | ๐ Error breakpoint analysis |
๐ Information context constraint testing | ๐ Anti-agent roadblocks (i.e. Captchas) | ๐ CI/CD integration checks | |
๐ Regression testing | ๐ Multi-agent framework visualization |
Our mission is to bring your agent from prototype to production.
Agent developers often work with little to no visibility into agent testing performance. This means their agents never leave the lab. We're changing that.
AgentOps is the easiest way to evaluate, grade, and test agents. Is there a feature you'd like to see AgentOps cover? Just raise it in the issues tab, and we'll work on adding it to the roadmap.