mirror of
https://github.com/hanzoai/insights.git
synced 2026-10-30 16:15:31 +00:00
zeekay
159e825022
The five days of upstream evolution since the 2026-08-01 import land through bin/debrand: new query schema, billing counters, ducklake compaction, the desktop sandbox default, capture evolution. Ours' invariants hold everywhere they met it: IAM-native auth only, no ee, no upsell surfaces, one sessions model, org-is-the-tenant, v1-first serving, fail-closed capture. Crates and workflows the fork retired stay retired; the Lemon prefix finishes dissolving into elements.
181 lines
7.6 KiB
TypeScript
181 lines
7.6 KiB
TypeScript
import { OnboardingComponentsContext, createInstallation } from 'scenes/onboarding/shared/OnboardingDocsContentWrapper'
|
|
|
|
import { StepDefinition } from '../steps'
|
|
|
|
export const getCrewAISteps = (ctx: OnboardingComponentsContext): StepDefinition[] => {
|
|
const { CodeBlock, CalloutBox, Markdown, dedent, snippets } = ctx
|
|
|
|
const NotableGenerationProperties = snippets?.NotableGenerationProperties
|
|
|
|
return [
|
|
{
|
|
title: 'Install the Insights SDK',
|
|
badge: 'required',
|
|
content: (
|
|
<>
|
|
<Markdown>
|
|
Setting up analytics starts with installing the Insights SDK. CrewAI uses LiteLLM under the hood,
|
|
and Insights integrates with LiteLLM's callback system.
|
|
</Markdown>
|
|
|
|
<CodeBlock
|
|
language="bash"
|
|
code={dedent`
|
|
pip install insights
|
|
`}
|
|
/>
|
|
</>
|
|
),
|
|
},
|
|
{
|
|
title: 'Install CrewAI',
|
|
badge: 'required',
|
|
content: (
|
|
<>
|
|
<Markdown>
|
|
Install CrewAI. Insights instruments your LLM calls through LiteLLM's callback system that CrewAI
|
|
uses natively.
|
|
</Markdown>
|
|
|
|
<CodeBlock
|
|
language="bash"
|
|
code={dedent`
|
|
pip install crewai litellm
|
|
`}
|
|
/>
|
|
</>
|
|
),
|
|
},
|
|
{
|
|
title: 'Configure Insights with LiteLLM',
|
|
badge: 'required',
|
|
content: (
|
|
<>
|
|
<Markdown>
|
|
Set your Insights project token and host as environment variables, then configure LiteLLM to use
|
|
Insights as a callback handler. You can find your project token in [your project
|
|
settings](https://app.hanzo.ai/settings/project).
|
|
</Markdown>
|
|
|
|
<CodeBlock
|
|
language="python"
|
|
code={dedent`
|
|
import os
|
|
import litellm
|
|
from crewai import Agent, Task, Crew, LLM
|
|
|
|
# Set Insights environment variables
|
|
os.environ["INSIGHTS_API_KEY"] = "<ph_project_token>"
|
|
os.environ["INSIGHTS_API_URL"] = "<ph_client_api_host>"
|
|
|
|
# Enable Insights callbacks in LiteLLM
|
|
litellm.success_callback = ["insights"]
|
|
litellm.failure_callback = ["insights"]
|
|
`}
|
|
/>
|
|
|
|
<CalloutBox type="fyi" icon="IconInfo" title="How this works">
|
|
<Markdown>
|
|
CrewAI can route LLM calls either through its own provider clients or through LiteLLM.
|
|
Insights hooks into LiteLLM's callback system, so you need `is_litellm=True` on the `LLM` you
|
|
pass to your agents. With it, Insights captures every call as an `$ai_generation` event,
|
|
without proxying your calls.
|
|
</Markdown>
|
|
</CalloutBox>
|
|
</>
|
|
),
|
|
},
|
|
{
|
|
title: 'Run your crew',
|
|
badge: 'required',
|
|
content: (
|
|
<>
|
|
<Markdown>
|
|
{dedent`
|
|
Run your CrewAI agents as normal. Insights automatically captures an \`$ai_generation\`
|
|
event for each LLM call. LiteLLM's callback does not see the tools your agents call.
|
|
Capture a tool's own execution as a span from inside the tool itself instead, as
|
|
\`my_tool\` does below.
|
|
`}
|
|
</Markdown>
|
|
|
|
<CodeBlock
|
|
language="python"
|
|
code={dedent`
|
|
from insights import Insights
|
|
from crewai.tools import tool
|
|
import time, uuid
|
|
|
|
insights = Insights("<ph_project_token>", host="<ph_client_api_host>")
|
|
|
|
trace_id = str(uuid.uuid4())
|
|
|
|
@tool
|
|
def my_tool(query: str) -> str:
|
|
"""Describe what your tool does."""
|
|
start = time.time()
|
|
result = run_tool(query)
|
|
|
|
insights.capture(
|
|
distinct_id="user_123",
|
|
event="$ai_span",
|
|
properties={
|
|
"$ai_trace_id": trace_id,
|
|
"$ai_session_id": "conversation-abc",
|
|
"$ai_span_id": str(uuid.uuid4()),
|
|
"$ai_span_name": "my_tool",
|
|
"$ai_input_state": {"query": query},
|
|
"$ai_output_state": result,
|
|
"$ai_latency": time.time() - start,
|
|
},
|
|
)
|
|
return result
|
|
|
|
# is_litellm=True routes calls through LiteLLM so the Insights
|
|
# callback fires. Without it, CrewAI uses its own provider client
|
|
# and no events are captured.
|
|
llm = LLM(
|
|
model="gpt-4o-mini",
|
|
is_litellm=True,
|
|
metadata={
|
|
"user_id": "user_123",
|
|
"$ai_session_id": "conversation-abc",
|
|
"$ai_trace_id": trace_id,
|
|
},
|
|
)
|
|
|
|
researcher = Agent(
|
|
role="Researcher",
|
|
goal="Find the weather in a city",
|
|
backstory="You are an expert wildlife researcher.",
|
|
llm=llm,
|
|
tools=[my_tool],
|
|
)
|
|
|
|
task = Task(
|
|
description="Find the weather in Paris.",
|
|
expected_output="The weather in Paris.",
|
|
agent=researcher,
|
|
)
|
|
|
|
crew = Crew(agents=[researcher], tasks=[task])
|
|
result = crew.kickoff()
|
|
print(result)
|
|
`}
|
|
/>
|
|
|
|
<Markdown>
|
|
{dedent`
|
|
You can expect captured \`$ai_generation\` events to have the following properties:
|
|
`}
|
|
</Markdown>
|
|
|
|
{NotableGenerationProperties && <NotableGenerationProperties />}
|
|
</>
|
|
),
|
|
},
|
|
]
|
|
}
|
|
|
|
export const CrewAIInstallation = createInstallation(getCrewAISteps)
|