You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+44Lines changed: 44 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -19,6 +19,50 @@ MASEval is an evaluation library that provides a unified interface for benchmark
19
19
20
20
Analogous to pytest for testing or MLflow for ML experimentation, MASEval focuses exclusively on evaluation infrastructure. It does not implement agents, define multi-agent communication protocols, or turn LLMs into agents. Instead, it wraps existing agent systems via simple adapters, orchestrates the evaluation lifecycle (setup, execution, measurement, teardown), and provides lifecycle hooks for tracing, logging, and metrics collection. This separation allows researchers to compare different agent architectures apples-to-apples across frameworks, while maintaining full control over their agent implementations.
21
21
22
+
## Why MASEval?
23
+
24
+
Compare multi-agent evaluation frameworks across key capabilities.
|**Multi-Agent**| Multi-Agent Native | Native orchestration with per-agent tracing, independent message histories, and explicit coordination patterns. |
52
+
|**System Evaluation**| System-Level Comparison | Compare different framework implementations on the same benchmark (not just swapping LLMs). |
53
+
|**Agent Agnostic**| Agent Framework Agnostic | Evaluate agents from any framework via thin adapters without requiring protocol adoption or code recreation. |
54
+
|**Benchmarks**| Pre-Implemented Benchmarks | Ships complete, ready-to-run benchmarks with environments, tools, and evaluators (not just templates). |
55
+
|**Multi-turn User**| User-Agent Multi-turn | First-class user simulation with personas, stop tokens, and tool access for realistic multi-turn conversations. |
56
+
|**No Lock-In**| No Vendor Lock-In | Fully open-source, works offline, permissive license (MIT/Apache), no mandatory cloud services or telemetry. |
57
+
|**BYO**| BYO Philosophy | Bring your own logging, agents, environments, and tools — flexibility over opinionated defaults. |
58
+
|**State-Action Eval**| Trace-First Evaluation | Evaluate intermediate steps and tool usage patterns via trace filtering, not just final output scoring. |
59
+
|**Error Attr**| Structured Error Attribution | Structured exceptions distinguish between different failure for fair scoring (`AgentError` vs `EnvironmentError`). |
60
+
|**Lightweight**| Lightweight | Minimal dependencies, small codebase (~20k LOC), quick time to first evaluation (~5-15 min). |
61
+
|**Project Maturity**| Professional Tooling | Published on PyPI, CI/CD, good test coverage, structured logging, active maintenance, excellent docs. |
62
+
|**Sandbox**| Sandboxed Execution | Built-in Docker/K8s/VM isolation for safe code execution (or BYO sandbox via abstract Environment). |
63
+
64
+
</details>
65
+
22
66
## Core Principles:
23
67
24
68
-**Evaluation, Not Implementation:** MASEval provides the evaluation infrastructure—you bring your agent implementation. Whether you've built agents with AutoGen, LangChain, custom code, or direct LLM calls, MASEval wraps them via simple adapters and runs them through standardized benchmarks.
| **Multi-Agent** | Multi-Agent Native | Native orchestration with per-agent tracing, independent message histories, and explicit coordination patterns. |
48
+
| **System Evaluation** | System-Level Comparison | Compare different framework implementations on the same benchmark (not just swapping LLMs). |
49
+
| **Agent Agnostic** | Agent Framework Agnostic | Evaluate agents from any framework via thin adapters without requiring protocol adoption or code recreation. |
50
+
| **Benchmarks** | Pre-Implemented Benchmarks | Ships complete, ready-to-run benchmarks with environments, tools, and evaluators (not just templates). |
51
+
| **Multi-turn User** | User-Agent Multi-turn | First-class user simulation with personas, stop tokens, and tool access for realistic multi-turn conversations. |
52
+
| **No Lock-In** | No Vendor Lock-In | Fully open-source, works offline, permissive license (MIT/Apache), no mandatory cloud services or telemetry. |
53
+
| **BYO** | BYO Philosophy | Bring your own logging, agents, environments, and tools — flexibility over opinionated defaults. |
54
+
| **State-Action Eval** | Trace-First Evaluation | Evaluate intermediate steps and tool usage patterns via trace filtering, not just final output scoring. |
55
+
| **Error Attr** | Structured Error Attribution | Structured exceptions distinguish between different failure for fair scoring (`AgentError` vs `EnvironmentError`). |
56
+
| **Lightweight** | Lightweight | Minimal dependencies, small codebase (~20k LOC), quick time to first evaluation (~5-15 min). |
57
+
| **Project Maturity** | Professional Tooling | Published on PyPI, CI/CD, good test coverage, structured logging, active maintenance, excellent docs. |
58
+
| **Sandbox** | Sandboxed Execution | Built-in Docker/K8s/VM isolation for safe code execution (or BYO sandbox via abstract Environment). |
59
+
19
60
## Core Principles
20
61
21
62
-**Evaluation, Not Implementation:** MASEval provides the evaluation infrastructure—you bring your agent implementation. Whether you've built agents with AutoGen, LangChain, custom code, or direct LLM calls, MASEval wraps them via simple adapters and runs them through standardized benchmarks.
@@ -34,20 +75,6 @@ More details in the [Quickstart](getting-started/quickstart.md)
34
75
35
76
-**Abstract Base Classes:** The library provides abstract base classes for core components (Task, Benchmark, Environment, Evaluator) with optional default implementations, giving users flexibility to customize while maintaining interface consistency.
36
77
37
-
## Quickstart
38
-
39
-
Install the package from PyPI:
40
-
41
-
```bash
42
-
pip install maseval
43
-
```
44
-
45
-
Run the example script shipped with the repository:
46
-
47
-
```bash
48
-
python examples/smolagents_research.py
49
-
```
50
-
51
78
## API
52
79
53
80
See the automatic API reference under `Reference`.
0 commit comments