aevals 0.1.0__tar.gz

aevals-0.1.0/.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,26 @@
+---
+name: Bug report
+labels: bug
+---
+
+**Describe the bug**
+A clear description of what's broken.
+
+**To reproduce**
+Steps to reproduce:
+1. ...
+2. ...
+
+**Expected behavior**
+What you expected to happen.
+
+**Environment**
+- OS:
+- Python version:
+- aevals version:
+- LLM SDK + version:
+
+**Logs / traceback**
+```
+paste here
+```

aevals-0.1.0/.github/ISSUE_TEMPLATE/feature_request.md

@@ -0,0 +1,13 @@
+---
+name: Feature request
+labels: enhancement
+---
+
+**Problem**
+What are you trying to do that isn't possible or is harder than it should be?
+
+**Proposed solution**
+How you'd like it to work.
+
+**Alternatives considered**
+Other approaches you've thought about.

aevals-0.1.0/.github/PULL_REQUEST_TEMPLATE.md

@@ -0,0 +1,13 @@
+## What
+
+Brief description of the change.
+
+## Why
+
+Motivation or issue reference.
+
+## Test plan
+
+- [ ] Tests pass (`pytest`)
+- [ ] Lint passes (`ruff check src/ tests/`)
+- [ ] Covered by existing or new tests

aevals-0.1.0/.github/workflows/ci.yml

@@ -0,0 +1,50 @@
+name: CI
+
+on:
+  push:
+    branches: [main]
+  pull_request:
+    branches: [main]
+  workflow_call:
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.12"]
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v5
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: pip install -e ".[dev]"
+
+      - name: Lint
+        run: ruff check src/ tests/
+
+      - name: Test
+        run: pytest --cov=aevals --cov-report=term-missing --cov-fail-under=90
+
+  lint:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install dependencies
+        run: pip install -e ".[dev]"
+
+      - name: Ruff check
+        run: ruff check src/ tests/
+
+      - name: Ruff format check
+        run: ruff format --check src/ tests/

aevals-0.1.0/.github/workflows/release.yml

@@ -0,0 +1,83 @@
+name: Release
+
+on:
+  push:
+    tags:
+      - "v*"
+
+permissions:
+  contents: write
+  id-token: write
+
+jobs:
+  # Gate: ensure CI passes before releasing
+  ci:
+    uses: ./.github/workflows/ci.yml
+
+  build:
+    needs: ci
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install build tools
+        run: pip install build
+
+      - name: Verify tag matches package version
+        run: |
+          TAG_VERSION="${GITHUB_REF#refs/tags/v}"
+          PKG_VERSION=$(python -c "import tomllib; print(tomllib.load(open('pyproject.toml','rb'))['project']['version'])")
+          if [ "$TAG_VERSION" != "$PKG_VERSION" ]; then
+            echo "::error::Tag version ($TAG_VERSION) does not match package version ($PKG_VERSION)"
+            exit 1
+          fi
+
+      - name: Build sdist and wheel
+        run: python -m build
+
+      - name: Upload dist artifacts
+        uses: actions/upload-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+  publish-pypi:
+    needs: build
+    runs-on: ubuntu-latest
+    environment: pypi
+    permissions:
+      id-token: write
+    steps:
+      - name: Download dist artifacts
+        uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+
+  github-release:
+    needs: build
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Download dist artifacts
+        uses: actions/download-artifact@v4
+        with:
+          name: dist
+          path: dist/
+
+      - name: Create GitHub Release
+        uses: softprops/action-gh-release@v2
+        with:
+          generate_release_notes: true
+          files: dist/*

aevals-0.1.0/.gitignore

@@ -0,0 +1,219 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#   Usually these files are written by a python script from a template
+#   before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+# Pipfile.lock
+
+# UV
+uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+# poetry.lock
+# poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+# pdm.lock
+# pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+# pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# Redis
+*.rdb
+*.aof
+*.pid
+
+# RabbitMQ
+mnesia/
+rabbitmq/
+rabbitmq-data/
+
+# ActiveMQ
+activemq-data/
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#   JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#   be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#   and can be added to the global gitignore or merged into this file.  For a more nuclear
+#   option (not recommended) you can uncomment the following to ignore the entire idea folder.
+# .idea/
+
+# Abstra
+#   Abstra is an AI-powered process automation framework.
+#   Ignore directories containing user credentials, local state, and settings.
+#   Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#   Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore
+#   that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#   and can be added to the global gitignore or merged into this file. However, if you prefer,
+#   you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/
+
+# Streamlit
+.streamlit/secrets.toml
+
+# OS
+.DS_Store
+
+# aevals local data (traces, results, wrapper)
+.aevals/

aevals-0.1.0/.pre-commit-config.yaml

@@ -0,0 +1,16 @@
+repos:
+  - repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.15.5
+    hooks:
+      - id: ruff
+        args: [--fix, --exit-non-zero-on-fix]
+      - id: ruff-format
+
+  - repo: local
+    hooks:
+      - id: pyright
+        name: pyright
+        entry: pyright src/
+        language: system
+        types: [python]
+        pass_filenames: false

aevals-0.1.0/CHANGELOG.md

@@ -0,0 +1,22 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+This changelog is automatically generated by [git-cliff](https://git-cliff.org/).
+
+## [0.1.0] - 2026-03-14
+
+### Added
+
+- Core eval framework: subprocess isolation, OTel trace capture, two-track evaluation
+- Five deterministic constraints: `max_duration_ms`, `max_steps`, `tool_sequence`, `no_repeat_calls`, `output_contains`
+- LLM-as-judge rubric evaluation via litellm
+- CLI commands: `aevals init`, `aevals run`, `aevals mcp-serve`
+- MCP server with five tools: scan, init, run, results, trajectory
+- AST-based SDK and entrypoint detection (scanner)
+- Auto-instrumentation for 6 LLM SDKs via OpenLLMetry
+- Rich terminal output and JSON output formats
+- CI workflow with 90% coverage threshold
+- Release workflow with PyPI publishing
+- Example agents: booking-agent, sdr-agent
+- Pre-commit hooks: ruff, pyright

aevals-0.1.0/CLAUDE.md

@@ -0,0 +1,74 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## Commands
+
+```bash
+# Install
+uv pip install -e ".[dev]"
+
+# Test
+pytest                                                    # all tests
+pytest tests/test_constraints.py                          # single file
+pytest tests/test_constraints.py::TestMaxDuration          # single class
+pytest tests/test_constraints.py::TestMaxDuration::test_passes_under_limit  # single test
+pytest --cov=aevals --cov-report=term-missing --cov-fail-under=90  # with coverage (CI threshold: 90%)
+
+# Lint & format
+ruff check src/ tests/
+ruff check --fix src/ tests/
+ruff format src/ tests/
+ruff format --check src/ tests/                           # CI format check (no changes)
+
+# Type check
+pyright
+
+# CLI
+aevals init                                               # scan project, generate aevals.yaml + .aevals/
+aevals run                                                # run all scenarios
+aevals run --scenario <name>                              # run specific scenario
+aevals run --json                                         # machine-readable output
+aevals mcp-serve                                          # start MCP server for Claude Code
+```
+
+## Architecture
+
+aevals is an eval framework for LLM-based agents. It runs agent code in subprocesses, captures OpenTelemetry traces, and evaluates results against deterministic constraints and LLM-judged rubrics.
+
+### Data flow
+
+```
+aevals.yaml → scenario selection → subprocess (run_wrapper.py)
+  → OTel instruments LLM calls → trace JSON file
+  → parse into Trajectory → evaluate constraints + rubric → ScenarioResult
+```
+
+### Key modules
+
+- **`config/`** — Pydantic models for `aevals.yaml`. `AevalsConfig.resolve_constraints()` merges defaults with scenario overrides.
+- **`capture/otel.py`** — `activate()` sets up OTel tracing in agent subprocess. `parse_trace_file()` converts OTel span JSON into `LLMCall` objects by reading `gen_ai.*` attributes. Auto-instruments 6 LLM SDKs.
+- **`scanner/`** — AST-based detection of SDK imports and entrypoint candidates. Generates `run_wrapper.py` template with marker constants from `constants.py`.
+- **`evals/constraints.py`** — Five deterministic checks (max_duration_ms, max_steps, tool_sequence, no_repeat_calls, output_contains). Zero LLM cost.
+- **`evals/judge.py`** — LLM-as-judge via litellm. Sends trajectory + rubric, parses JSON response with index-based matching and positional fallback.
+- **`evals/runner.py`** — Shared `filter_scenarios()` and `summarize_results()` used by both CLI and MCP server.
+- **`cli/run.py`** — Spawns agent subprocesses concurrently via `asyncio.create_subprocess_exec`. Output extracted between `RESULT_MARKER_START/END` sentinels.
+- **`mcp/server.py`** — Primary interface for Claude Code. Five tools: scan, init, run, results, trajectory.
+- **`constants.py`** — Single source of truth for directory paths and output marker strings.
+
+### Subprocess isolation model
+
+Each scenario runs in its own subprocess with env vars (`AEVALS_RUN_ID`, `AEVALS_ENTRY`, `AEVALS_TRACE_DIR`). The generated `run_wrapper.py` dynamically imports the agent entry point, pipes input via stdin, and captures output between marker strings in stdout. OTel spans are written to per-run JSON files in `.aevals/traces/`.
+
+### Two-track evaluation
+
+Pass/fail is `constraints_pass AND rubric_pass`. Constraints are synchronous and deterministic. Rubric evaluation is async via litellm, optional (missing judge config → rubric_pending=True, doesn't fail). Results are Pydantic models throughout.
+
+## Conventions
+
+- Python 3.12+. Line length 100. Ruff rules: E, F, I, UP, B, SIM, RUF, PT, PIE, C4, RET, PERF.
+- Pyright strict mode. Type hints everywhere.
+- Async tests use `pytest-asyncio` with `asyncio_mode = "auto"` (no `@pytest.mark.asyncio` needed).
+- All shared path constants live in `constants.py` — don't hardcode `.aevals/` paths.
+- Output marker strings (`__AEVALS_RESULT__`, `__AEVALS_END__`) are defined once in `constants.py` and injected into the wrapper template via `.format()`.
+- TOCTOU: use `try/except FileNotFoundError` instead of `if path.exists()` then `path.read_text()`.

aevals-0.1.0/CONTRIBUTING.md

@@ -0,0 +1,62 @@
+# Contributing to aevals
+
+Thanks for your interest in contributing. This guide covers setup, workflow, and conventions.
+
+## Setup
+
+```bash
+git clone https://github.com/satyaborg/aevals.git
+cd aevals
+pip install -e ".[dev]"
+pre-commit install
+```
+
+## Development workflow
+
+1. Create a branch: `git checkout -b type/short-description`
+   - Types: `feat/`, `fix/`, `chore/`, `docs/`, `test/`
+2. Make your changes
+3. Run checks:
+
+```bash
+pytest                                      # tests
+pytest --cov=aevals --cov-fail-under=90     # with coverage
+ruff check src/ tests/                      # lint
+ruff format src/ tests/                     # format
+```
+
+4. Commit with a clear, imperative message under 72 chars
+5. Open a PR against `main`
+
+## Code conventions
+
+- **Python 3.12+**. Line length 100.
+- **Type hints everywhere**. No `Any` unless forced by a library boundary. Strict pyright.
+- **Double quotes** for strings (ruff default). Trailing commas always.
+- **Imports**: stdlib, third-party, local — separated by blank lines. Sorted by ruff/isort.
+- **Shared constants** live in `constants.py`. Don't hardcode `.aevals/` paths.
+- **File access**: use `try/except FileNotFoundError` instead of `if path.exists()` then read (TOCTOU).
+- **Async tests** use `pytest-asyncio` with `asyncio_mode = "auto"` — no `@pytest.mark.asyncio` needed.
+
+## Testing
+
+- Write tests for any function with branching logic or >10 lines.
+- Prefer real objects over mocks. Mock only at system boundaries (network, disk, time).
+- Test names: `test_<what>_<condition>_<expected>`.
+- CI enforces 90% coverage.
+
+## Pull requests
+
+- Keep PRs small and focused. If a PR touches >300 lines, consider splitting it.
+- One logical change per commit.
+- PRs run CI automatically (lint + test).
+
+## Lint rules
+
+Ruff is configured with: `E`, `F`, `I`, `UP`, `B`, `SIM`, `RUF`, `PT`, `PIE`, `C4`, `RET`, `PERF`.
+
+Pre-commit hooks run ruff and pyright automatically on commit.
+
+## Releases
+
+Releases are automated via GitHub Actions when a version tag is pushed. See `scripts/release.sh`.

aevals-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Satya Borgohain
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.