tracerazor-crewai

CrewAI callback adapter for TraceRazor.

Automatically captures every task execution and tool call from your CrewAI crew with zero manual instrumentation.

v0.2.0 — New Metrics:

• ✨ Semantic Continuity (CSD) — Detects when your crew's agents drift topic mid-workflow
• ✨ Adherence Scoring (IAR) — After optimizing, validates that fixes actually improved metrics

Install

pip install tracerazor-crewai
pip install tracerazor-crewai[crewai]  # includes crewai

Requires the tracerazor binary:

cargo build --release
export TRACERAZOR_BIN=/path/to/TraceRazor/target/release/tracerazor

Usage

from tracerazor_crewai import TraceRazorCallback
from crewai import Agent, Task, Crew

callback = TraceRazorCallback(
    agent_name="support-crew",
    threshold=70,
)

crew = Crew(
    agents=[...],
    tasks=[...],
    callbacks=[callback],
)

crew.kickoff()

# After the crew finishes:
report = callback.analyse()
print(report.markdown())

# CI/CD gate — raises AssertionError if TAS < threshold:
callback.assert_passes()

API

TraceRazorCallback(agent_name, framework, threshold, task_value_score, tracerazor_bin)

callback.analyse() → TraceRazorReport

Finalise and submit the trace. Returns the report.

callback.assert_passes()

Raise AssertionError if TAS < threshold.

callback.set_task_value_score(score: float)

Update quality score before calling analyse().

Captured events

Multi-Agent Crews

For complex workflows with multiple crews, use a separate callback for each:

from tracerazor_crewai import TraceRazorCallback

# Crew 1: Triage & classification
triage_crew = Crew(agents=[triage_agent], tasks=[triage_task], 
                   callbacks=[TraceRazorCallback(agent_name="triage-crew")])
triage_result = triage_crew.kickoff()

# Crew 2: Resolution execution
resolution_crew = Crew(agents=[resolution_agent], tasks=[resolution_task],
                       callbacks=[TraceRazorCallback(agent_name="resolution-crew")])
resolution_result = resolution_crew.kickoff()

# Each crew is audited independently

See Multi-Agent Guide for complete example with per-agent optimization.