Multi-Agent Eval Harness

Overview

Eval that works for a single agent does not catch coordination failures: dropped handoffs, role confusion, blame loops, partial state transfer. This harness adds role-level metrics, coordination-failure scenarios, and full transcript grading.

When to use

≥2 agents collaborating (orchestrator + workers, peer agents, A2A)
Bugs that only appear in production multi-agent runs
Need to compare orchestration patterns (hierarchical vs peer vs blackboard)
A2A protocol integration (see a2a-protocol)

What to measure

Layer	Metric	Why
Per-role	task completion rate	Did each role do its job?
Per-role	tool-call validity	Did each role use tools correctly?
Handoff	state-transfer completeness	Did receiver get what it needs?
Handoff	handoff latency	Coordination overhead
System	end-to-end success	Did the whole thing work?
System	total cost / latency	Production viability
Failure	blame loop detection	Are agents finger-pointing?
Failure	deadlock detection	Waiting on each other?

Coordination-failure test suite

Inject these scenarios deliberately:

Drop a message — verify receiver detects, retries
Stale state — receiver gets outdated context, must refresh
Conflicting instructions — two agents send opposing orders, who wins?
Silent peer failure — peer agent times out without error
Role bleed — orchestrator tries to do worker's job
Authority confusion — worker overrides orchestrator decision
Budget exhaustion mid-handoff — graceful or stuck?

Transcript grading

RUBRIC = {
    "role_adherence": "Did each agent stay in its declared role?",
    "handoff_quality": "Did handoffs include needed context?",
    "termination": "Did the system stop cleanly when done?",
    "blame_loop": "Score 0 if any blame loop detected.",
    "user_facing_quality": "Final answer quality vs golden.",
}

def grade_transcript(transcript, golden):
    return judge_llm.score(transcript, rubric=RUBRIC, golden=golden)

LLM-as-judge prompt must see the full transcript with role labels, not just final output.

Harness layout

multi_agent_eval/
  scenarios/
    drop_message.yaml
    stale_state.yaml
    ...
  runs/
    <ts>/
      transcript.jsonl
      scores.json
  grader.py
  reporter.py

Quick start

from harness import Scenario, run_scenario, grade

scenarios = Scenario.load_all("scenarios/")
results = [run_scenario(s, agents=my_system) for s in scenarios]
scores = [grade(r, rubric=RUBRIC) for r in results]
report(scores)

CI gate

Block merge if role_adherence < 0.9 on regression set
Block if any new blame_loop detected
Block if end-to-end success drops > 5pp vs baseline

multi-agent-eval-harness