agent-trajectory-diff 0.1.0__tar.gz → 0.2.0__tar.gz

agent_trajectory_diff-0.2.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 AgentDiff contributors
+
+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.

agent_trajectory_diff-0.2.0/PKG-INFO

@@ -0,0 +1,204 @@
+Metadata-Version: 2.5
+Name: agent-trajectory-diff
+Version: 0.2.0
+Summary: A developer-first package to evaluate and find regressions in agent trajectories
+Project-URL: Homepage, https://github.com/lostmartian/agentdiff
+Project-URL: Repository, https://github.com/lostmartian/agentdiff
+Author-email: Sahil Gangurde <sahilgangurde08@gmail.com>
+License-Expression: MIT
+License-File: LICENSE
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Requires-Python: >=3.10
+Requires-Dist: networkx>=3.0
+Requires-Dist: pydantic>=2.0
+Requires-Dist: rich>=13.0
+Requires-Dist: tomli>=2.0; python_version < '3.11'
+Requires-Dist: typer>=0.9
+Description-Content-Type: text/markdown
+
+# AgentDiff
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![Python Version](https://img.shields.io/badge/python-3.10%2B-blue.svg)](pyproject.toml)
+
+**AgentDiff** is a developer-first Python library and CLI designed to solve the hardest problem in agent engineering: **regression testing multi-turn, tool-using AI agents by comparing execution paths (trajectories) head-to-head.**
+
+## What AgentDiff Is
+
+* **A Trajectory Diff Engine:** Compares Run A (Baseline) against Run B (Candidate) across their execution Directed Acyclic Graphs (DAGs).
+* **A Local-First CI/CD Gate:** Runs locally in your terminal or inside `pytest` and GitHub Actions, raising errors or exit codes on regression violations.
+* **A Universal Comparator:** Ingests telemetry run files from **OpenInference/OTel**, **Langfuse**, **LangSmith**, **OpenAI Agents SDK**, or raw/custom JSON.
+
+## Installation
+
+Install the PyPI package:
+```bash
+pip install agent-trajectory-diff
+```
+
+Or using `uv`:
+```bash
+uv add agent-trajectory-diff
+```
+
+## Quickstart
+
+### 1. CLI Usage
+
+Compare two trajectory JSON traces from your terminal:
+```bash
+agentdiff baseline_run.json candidate_run.json --fail-on-regression --max-divergence 0.25
+```
+
+Options:
+- `--adapter`: Telemetry parser to use (`auto`, `generic`, `openinference`, `langfuse`, `langsmith`, `openai_agents`).
+- `--format`: Format for the output (`terminal`, `json`, `markdown`).
+- `--fail-on-regression`: Return exit code `1` if thresholds are violated.
+- `--max-loops`: Maximum loops allowed.
+- `--max-divergence`: Maximum Trajectory Divergence Index (TDI) allowed.
+- `--max-cost-delta`: Maximum cost increase percentage allowed.
+- `--baseline, -b PATH`: Compare against a persistent baseline trace file (see [Baseline workflow](#baseline-workflow)).
+- `--update-baseline`: Overwrite the persistent baseline with the candidate after a clean diff.
+- `--config PATH`: Load defaults from an `agentdiff.toml` (auto-discovered if not given).
+
+#### Config-as-code (`agentdiff.toml`)
+
+Commit your thresholds, adapter, and baseline path next to your traces instead of repeating CLI flags. Explicit flags always win over config.
+
+```toml
+[compare]
+detect_loops = true
+strict_tool_signatures = false
+
+[adapter]
+name = "auto"            # auto, generic, openinference, langfuse, langsmith, openai_agents
+
+[cli]
+format = "terminal"      # terminal, json, markdown, pr
+baseline = "baselines/current.json"
+max_loops = 0
+max_divergence = 0.3
+max_cost_delta = 10.0
+
+[assertions]             # defaults used by assert_no_regressions / pytest plugin
+max_divergence = 0.25
+max_cost_increase_pct = 5.0
+allow_loops = false
+max_wasted_effort = 0.1
+```
+
+AgentDiff auto-discovers `agentdiff.toml` from the current directory upward, or you can point at it explicitly with `--config`.
+
+## Baseline workflow
+
+Keep a single `baseline.json` file committed to your repo instead of hand-managing two trace files. The first run establishes the baseline; later runs compare against it and advance it only on clean diffs.
+
+```bash
+# First run: stores candidate as the baseline, exits 0
+agentdiff baseline.json today.json --baseline baseline.json --update-baseline
+
+# Later runs: compare today's run against the stored baseline
+agentdiff baseline.json today.json --baseline baseline.json --update-baseline --fail-on-regression
+```
+
+- If `baseline.json` does not exist and `--update-baseline` is set, the candidate is copied in as the baseline and the command exits `0`.
+- If it does not exist and `--update-baseline` is omitted, the command exits `2` with a helpful message.
+- On a regression the baseline is **never** overwritten, and `--fail-on-regression` exits `1`.
+
+### 2. Python SDK & Pytest Integration
+
+Catch agent loop regressions or token cost spikes in your test suites:
+
+```python
+import pytest
+from agentdiff import load_trace, compare
+from agentdiff.testing import assert_no_regressions
+
+def test_agent_refactor_efficiency():
+    # Load traces from disk (auto-detects the telemetry format)
+    baseline = load_trace("tests/traces/baseline.json")
+    candidate = load_trace("tests/traces/candidate.json")
+
+    # Run the comparison
+    report = compare(baseline, candidate)
+
+    # Expressive assertion helper that raises detailed error messages on regression
+    assert_no_regressions(
+        report,
+        max_divergence=0.25,        # TDI threshold [0.0 - 1.0]
+        max_cost_increase_pct=5.0,  # Max cost increase allowed
+        allow_loops=False,           # Reject if tool loops are detected
+        max_wasted_effort=0.10      # Max Wasted Effort Index (WEI) allowed
+    )
+```
+
+### 3. GitHub Action
+
+Gate a PR on agent trajectory regressions with the reusable composite action.
+Pin it to a release tag and point `package` at the published package (or a
+`git+` path / local directory for pre-release testing):
+
+```yaml
+name: AgentDiff Gate
+on:
+  pull_request:
+
+jobs:
+  agentdiff:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - uses: lostmartian/agentdiff/.github/actions/agentdiff-check@v0.2.0
+        with:
+          baseline: traces/baseline.json   # committed baseline trace
+          candidate: traces/candidate.json # generated by an earlier step
+          update-baseline: "false"
+          max-divergence: "0.3"
+          max-cost-delta: "10.0"
+```
+
+The action installs the package (default `agent-trajectory-diff` from PyPI),
+runs `agentdiff --fail-on-regression`, and fails the job when divergence,
+loops, or cost spikes exceed the thresholds.
+
+**Available inputs:**
+
+| Input | Default | Description |
+| --- | --- | --- |
+| `baseline` | *(required)* | Path to the stored baseline trace JSON. |
+| `candidate` | *(required)* | Path to the candidate trace JSON. |
+| `package` | `agent-trajectory-diff` | Python package spec to install (PyPI name, `git+https://…`, or a local path). |
+| `adapter` | `auto` | Telemetry adapter: `auto`, `generic`, `openinference`, `langfuse`, `langsmith`, `openai_agents`. |
+| `max-divergence` | `0.3` | Maximum Trajectory Divergence Index (TDI) before regression. |
+| `max-loops` | `0` | Maximum loop count before regression. |
+| `max-cost-delta` | `10.0` | Maximum cost increase percentage before regression. |
+| `update-baseline` | `false` | Overwrite the stored baseline with the candidate when the run is clean. |
+
+## Core Metrics
+
+| Metric | Target / Range | Algorithmic Definition |
+| --- | --- | --- |
+| **Trajectory Divergence Index (TDI)** | `0.0` (Identical) to `1.0` (Divergent) | $$1.0 - \frac{2 \times \vert{}\text{LCS}(\text{Steps}_A, \text{Steps}_B)\vert{}}{\vert{}\text{Steps}_A\vert{} + \vert{}\text{Steps}_B\vert{}}$$ |
+| **Wasted Effort Index (WEI)** | `0.0` (Optimal) to `1.0` (Total Waste) | $$\frac{\text{Count}(\text{Steps with status} \in \{\text{ERROR, RETRY, ABANDONED}\})}{\text{Total Execution Steps}}$$ |
+| **Loop Buster Index (LBI)** | Integer ($\ge 0$) | Detects consecutive repeating sequences of tools with stagnant state changes. |
+| **Resource Deltas ($\Delta\text{Res}$)** | Percentage ($\pm\%$) | Standard deltas for $\Delta\text{Tokens}$, $\Delta\text{Cost}$, and $\Delta\text{Latency}$. |
+
+## Development & Operations
+
+This project utilizes `uv` to manage environments and dependencies. Automation tasks are defined in the **[`Makefile`](file:///Users/lostmartian/Desktop/interview/agentdiff/Makefile)**:
+
+- `make lint` / `make format`: Run Ruff linter checks and formatter.
+- `make test`: Run pytest suite (including style & formatting assertions).
+- `make build`: Package the library into source and wheel distributions in `dist/`.
+- `make website-dev`: Start the Next.js landing and documentation site local server.
+- `make website-build`: Build the Next.js static output in `website/out/`.
+
+### Repository Layout
+- `src/`: Python source code package modules.
+- `tests/`: Quality assurance unit tests.
+- `website/`: Next.js web application and documentation pages.

agent_trajectory_diff-0.2.0/README.md

@@ -0,0 +1,184 @@
+# AgentDiff
+
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)
+[![Python Version](https://img.shields.io/badge/python-3.10%2B-blue.svg)](pyproject.toml)
+
+**AgentDiff** is a developer-first Python library and CLI designed to solve the hardest problem in agent engineering: **regression testing multi-turn, tool-using AI agents by comparing execution paths (trajectories) head-to-head.**
+
+## What AgentDiff Is
+
+* **A Trajectory Diff Engine:** Compares Run A (Baseline) against Run B (Candidate) across their execution Directed Acyclic Graphs (DAGs).
+* **A Local-First CI/CD Gate:** Runs locally in your terminal or inside `pytest` and GitHub Actions, raising errors or exit codes on regression violations.
+* **A Universal Comparator:** Ingests telemetry run files from **OpenInference/OTel**, **Langfuse**, **LangSmith**, **OpenAI Agents SDK**, or raw/custom JSON.
+
+## Installation
+
+Install the PyPI package:
+```bash
+pip install agent-trajectory-diff
+```
+
+Or using `uv`:
+```bash
+uv add agent-trajectory-diff
+```
+
+## Quickstart
+
+### 1. CLI Usage
+
+Compare two trajectory JSON traces from your terminal:
+```bash
+agentdiff baseline_run.json candidate_run.json --fail-on-regression --max-divergence 0.25
+```
+
+Options:
+- `--adapter`: Telemetry parser to use (`auto`, `generic`, `openinference`, `langfuse`, `langsmith`, `openai_agents`).
+- `--format`: Format for the output (`terminal`, `json`, `markdown`).
+- `--fail-on-regression`: Return exit code `1` if thresholds are violated.
+- `--max-loops`: Maximum loops allowed.
+- `--max-divergence`: Maximum Trajectory Divergence Index (TDI) allowed.
+- `--max-cost-delta`: Maximum cost increase percentage allowed.
+- `--baseline, -b PATH`: Compare against a persistent baseline trace file (see [Baseline workflow](#baseline-workflow)).
+- `--update-baseline`: Overwrite the persistent baseline with the candidate after a clean diff.
+- `--config PATH`: Load defaults from an `agentdiff.toml` (auto-discovered if not given).
+
+#### Config-as-code (`agentdiff.toml`)
+
+Commit your thresholds, adapter, and baseline path next to your traces instead of repeating CLI flags. Explicit flags always win over config.
+
+```toml
+[compare]
+detect_loops = true
+strict_tool_signatures = false
+
+[adapter]
+name = "auto"            # auto, generic, openinference, langfuse, langsmith, openai_agents
+
+[cli]
+format = "terminal"      # terminal, json, markdown, pr
+baseline = "baselines/current.json"
+max_loops = 0
+max_divergence = 0.3
+max_cost_delta = 10.0
+
+[assertions]             # defaults used by assert_no_regressions / pytest plugin
+max_divergence = 0.25
+max_cost_increase_pct = 5.0
+allow_loops = false
+max_wasted_effort = 0.1
+```
+
+AgentDiff auto-discovers `agentdiff.toml` from the current directory upward, or you can point at it explicitly with `--config`.
+
+## Baseline workflow
+
+Keep a single `baseline.json` file committed to your repo instead of hand-managing two trace files. The first run establishes the baseline; later runs compare against it and advance it only on clean diffs.
+
+```bash
+# First run: stores candidate as the baseline, exits 0
+agentdiff baseline.json today.json --baseline baseline.json --update-baseline
+
+# Later runs: compare today's run against the stored baseline
+agentdiff baseline.json today.json --baseline baseline.json --update-baseline --fail-on-regression
+```
+
+- If `baseline.json` does not exist and `--update-baseline` is set, the candidate is copied in as the baseline and the command exits `0`.
+- If it does not exist and `--update-baseline` is omitted, the command exits `2` with a helpful message.
+- On a regression the baseline is **never** overwritten, and `--fail-on-regression` exits `1`.
+
+### 2. Python SDK & Pytest Integration
+
+Catch agent loop regressions or token cost spikes in your test suites:
+
+```python
+import pytest
+from agentdiff import load_trace, compare
+from agentdiff.testing import assert_no_regressions
+
+def test_agent_refactor_efficiency():
+    # Load traces from disk (auto-detects the telemetry format)
+    baseline = load_trace("tests/traces/baseline.json")
+    candidate = load_trace("tests/traces/candidate.json")
+
+    # Run the comparison
+    report = compare(baseline, candidate)
+
+    # Expressive assertion helper that raises detailed error messages on regression
+    assert_no_regressions(
+        report,
+        max_divergence=0.25,        # TDI threshold [0.0 - 1.0]
+        max_cost_increase_pct=5.0,  # Max cost increase allowed
+        allow_loops=False,           # Reject if tool loops are detected
+        max_wasted_effort=0.10      # Max Wasted Effort Index (WEI) allowed
+    )
+```
+
+### 3. GitHub Action
+
+Gate a PR on agent trajectory regressions with the reusable composite action.
+Pin it to a release tag and point `package` at the published package (or a
+`git+` path / local directory for pre-release testing):
+
+```yaml
+name: AgentDiff Gate
+on:
+  pull_request:
+
+jobs:
+  agentdiff:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+      - uses: lostmartian/agentdiff/.github/actions/agentdiff-check@v0.2.0
+        with:
+          baseline: traces/baseline.json   # committed baseline trace
+          candidate: traces/candidate.json # generated by an earlier step
+          update-baseline: "false"
+          max-divergence: "0.3"
+          max-cost-delta: "10.0"
+```
+
+The action installs the package (default `agent-trajectory-diff` from PyPI),
+runs `agentdiff --fail-on-regression`, and fails the job when divergence,
+loops, or cost spikes exceed the thresholds.
+
+**Available inputs:**
+
+| Input | Default | Description |
+| --- | --- | --- |
+| `baseline` | *(required)* | Path to the stored baseline trace JSON. |
+| `candidate` | *(required)* | Path to the candidate trace JSON. |
+| `package` | `agent-trajectory-diff` | Python package spec to install (PyPI name, `git+https://…`, or a local path). |
+| `adapter` | `auto` | Telemetry adapter: `auto`, `generic`, `openinference`, `langfuse`, `langsmith`, `openai_agents`. |
+| `max-divergence` | `0.3` | Maximum Trajectory Divergence Index (TDI) before regression. |
+| `max-loops` | `0` | Maximum loop count before regression. |
+| `max-cost-delta` | `10.0` | Maximum cost increase percentage before regression. |
+| `update-baseline` | `false` | Overwrite the stored baseline with the candidate when the run is clean. |
+
+## Core Metrics
+
+| Metric | Target / Range | Algorithmic Definition |
+| --- | --- | --- |
+| **Trajectory Divergence Index (TDI)** | `0.0` (Identical) to `1.0` (Divergent) | $$1.0 - \frac{2 \times \vert{}\text{LCS}(\text{Steps}_A, \text{Steps}_B)\vert{}}{\vert{}\text{Steps}_A\vert{} + \vert{}\text{Steps}_B\vert{}}$$ |
+| **Wasted Effort Index (WEI)** | `0.0` (Optimal) to `1.0` (Total Waste) | $$\frac{\text{Count}(\text{Steps with status} \in \{\text{ERROR, RETRY, ABANDONED}\})}{\text{Total Execution Steps}}$$ |
+| **Loop Buster Index (LBI)** | Integer ($\ge 0$) | Detects consecutive repeating sequences of tools with stagnant state changes. |
+| **Resource Deltas ($\Delta\text{Res}$)** | Percentage ($\pm\%$) | Standard deltas for $\Delta\text{Tokens}$, $\Delta\text{Cost}$, and $\Delta\text{Latency}$. |
+
+## Development & Operations
+
+This project utilizes `uv` to manage environments and dependencies. Automation tasks are defined in the **[`Makefile`](file:///Users/lostmartian/Desktop/interview/agentdiff/Makefile)**:
+
+- `make lint` / `make format`: Run Ruff linter checks and formatter.
+- `make test`: Run pytest suite (including style & formatting assertions).
+- `make build`: Package the library into source and wheel distributions in `dist/`.
+- `make website-dev`: Start the Next.js landing and documentation site local server.
+- `make website-build`: Build the Next.js static output in `website/out/`.
+
+### Repository Layout
+- `src/`: Python source code package modules.
+- `tests/`: Quality assurance unit tests.
+- `website/`: Next.js web application and documentation pages.

{agent_trajectory_diff-0.1.0 → agent_trajectory_diff-0.2.0}/pyproject.toml

@@ -1,22 +1,22 @@
 [project]
 name = "agent-trajectory-diff"
-version = "0.1.0"
 description = "A developer-first package to evaluate and find regressions in agent trajectories"
 readme = "README.md"
+dynamic = ["version"]
 authors = [
     { name = "Sahil Gangurde", email = "sahilgangurde08@gmail.com" }
 ]
 requires-python = ">=3.10"
-license = { text = "Apache-2.0 OR MIT" }
+license = "MIT"
 dependencies = [
     "pydantic>=2.0",
     "networkx>=3.0",
     "typer>=0.9",
     "rich>=13.0",
+    "tomli>=2.0; python_version < '3.11'",
 ]
 classifiers = [
     "Programming Language :: Python :: 3",
-    "License :: OSI Approved :: Apache Software License",
     "License :: OSI Approved :: MIT License",
     "Operating System :: OS Independent",
 ]
@@ -24,13 +24,18 @@ classifiers = [
 [project.scripts]
 agentdiff = "agentdiff.cli:main"
 
+[project.entry-points.pytest11]
+agentdiff = "agentdiff.pytest_plugin"
+
 [project.urls]
 Homepage = "https://github.com/lostmartian/agentdiff"
 Repository = "https://github.com/lostmartian/agentdiff"
 
 [dependency-groups]
 dev = [
+    "hypothesis>=6.165.10",
     "pytest>=8.0",
+    "pytest-cov>=7.1.0",
     "ruff>=0.16.3",
 ]
 
@@ -38,6 +43,9 @@ dev = [
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
+[tool.hatch.version]
+path = "src/agentdiff/__init__.py"
+
 [tool.hatch.build.targets.sdist]
 include = [
     "/src",
@@ -66,3 +74,7 @@ ignore = ["E501"]
 [tool.ruff.format]
 quote-style = "double"
 indent-style = "space"
+
+[tool.pytest.ini_options]
+addopts = "-p pytester"
+testpaths = ["tests"]

agent_trajectory_diff-0.2.0/src/agentdiff/__init__.py

@@ -0,0 +1,93 @@
+"""AgentDiff — regression testing for multi-turn, tool-using AI agents.
+
+Compare two agent execution trajectories (the DAG of tool calls, LLM calls, and
+decisions) head-to-head to catch regressions in CI.
+
+Public API
+----------
+- ``compare(baseline, candidate)`` -> ``DiffReport``
+    Structural diff (alignment + TDI/WEI + resource deltas + loop detection).
+- ``assert_no_regressions(report, ...)``
+    Threshold gates (divergence, cost, loops, wasted effort) for pytest.
+- ``load_trace(path, adapter="auto")`` -> ``AgentTrace``
+    Load a trace from JSON, auto-detecting the telemetry format.
+- ``parse_trace_data(data, adapter="auto")`` -> ``AgentTrace``
+    Parse an in-memory dict instead of a file.
+- Adapters: ``GenericAdapter``, ``LangfuseAdapter``, ``LangSmithAdapter``,
+    ``OpenInferenceAdapter``, ``OpenAIAgentsAdapter`` (all expose
+    ``from_dict`` / ``from_file``).
+- ``load_config(path=None)`` -> ``AgentDiffConfig``
+    Load defaults from ``agentdiff.toml`` (thresholds, adapter, baseline).
+- Models: ``AgentTrace`` (canonical, ``schema_version``-ed), ``DiffReport``,
+    ``StepDiff``/``StepDiffStatus``, ``TraceStep``/``StepStatus``/``StepType``.
+
+Typical flow::
+
+    baseline = load_trace("baseline.json")
+    candidate = load_trace("candidate.json")
+    report = compare(baseline, candidate)
+    assert_no_regressions(report)
+"""
+
+__version__ = "0.2.0"
+
+from agentdiff.adapters import (
+    BaseAdapter,
+    GenericAdapter,
+    LangfuseAdapter,
+    LangSmithAdapter,
+    OpenAIAgentsAdapter,
+    OpenInferenceAdapter,
+)
+from agentdiff.ci.baseline import decide_rotation
+from agentdiff.config import AgentDiffConfig, load_config
+from agentdiff.engine import compare
+from agentdiff.engine.explanations import (
+    format_explanations,
+    generate_explanations,
+    locate_culprit,
+)
+from agentdiff.engine.tree import render_tree
+from agentdiff.loader import load_trace, parse_trace_data
+from agentdiff.models import (
+    AgentTrace,
+    DiffReport,
+    StepDiff,
+    StepDiffStatus,
+    StepStatus,
+    StepType,
+    TokenUsage,
+    TraceStep,
+)
+from agentdiff.reporters.pr import generate_pr_markdown
+from agentdiff.testing import assert_no_regressions
+
+__all__ = [
+    "AgentDiffConfig",
+    "AgentTrace",
+    "BaseAdapter",
+    "DiffReport",
+    "GenericAdapter",
+    "LangSmithAdapter",
+    "LangfuseAdapter",
+    "OpenAIAgentsAdapter",
+    "OpenInferenceAdapter",
+    "StepDiff",
+    "StepDiffStatus",
+    "StepStatus",
+    "StepType",
+    "TokenUsage",
+    "TraceStep",
+    "__version__",
+    "assert_no_regressions",
+    "compare",
+    "decide_rotation",
+    "format_explanations",
+    "generate_explanations",
+    "generate_pr_markdown",
+    "load_config",
+    "load_trace",
+    "locate_culprit",
+    "parse_trace_data",
+    "render_tree",
+]

{agent_trajectory_diff-0.1.0 → agent_trajectory_diff-0.2.0}/src/agentdiff/adapters/__init__.py

@@ -1,13 +1,15 @@
 from agentdiff.adapters.base import BaseAdapter
-from agentdiff.adapters.deepeval import DeepEvalAdapter
 from agentdiff.adapters.generic import GenericAdapter
 from agentdiff.adapters.langfuse import LangfuseAdapter
+from agentdiff.adapters.langsmith import LangSmithAdapter
+from agentdiff.adapters.openai_agents import OpenAIAgentsAdapter
 from agentdiff.adapters.openinference import OpenInferenceAdapter
 
 __all__ = [
     "BaseAdapter",
-    "DeepEvalAdapter",
     "GenericAdapter",
+    "LangSmithAdapter",
     "LangfuseAdapter",
+    "OpenAIAgentsAdapter",
     "OpenInferenceAdapter",
 ]

agent_trajectory_diff-0.2.0/src/agentdiff/adapters/_iso.py

@@ -0,0 +1,31 @@
+"""Robust ISO-8601 timestamp parsing shared across telemetry adapters.
+
+Python's ``datetime.fromisoformat`` is strict about fractional seconds on
+Python 3.10 (it requires exactly 6 digits). Telemetry timestamps commonly use
+a short fractional part (e.g. ``...00.5Z``) or ``Z``/``+00:00`` offsets. This
+helper normalizes both so latency math is identical across Python 3.10+.
+"""
+
+from __future__ import annotations
+
+import re
+from datetime import datetime
+from typing import Any
+
+_FRACTION = re.compile(r"\.(\d+)")
+
+
+def parse_iso_timestamp(value: Any) -> datetime | None:
+    """Parse a telemetry timestamp to an aware ``datetime``.
+
+    Returns ``None`` (never raises) when the value cannot be parsed.
+    """
+    if value is None:
+        return None
+    text = str(value).replace("Z", "+00:00")
+    if "." in text:
+        text = _FRACTION.sub(lambda m: "." + m.group(1).ljust(6, "0"), text, count=1)
+    try:
+        return datetime.fromisoformat(text)
+    except (ValueError, TypeError):
+        return None

agent_trajectory_diff-0.2.0/src/agentdiff/adapters/base.py

@@ -0,0 +1,42 @@
+import json
+from abc import ABC, abstractmethod
+from typing import Any
+
+from agentdiff.models.trace import AgentTrace
+
+
+class BaseAdapter(ABC):
+    """Converts a provider-specific trace format into a canonical :class:`AgentTrace`.
+
+    Subclasses implement :meth:`from_dict`, mapping a raw provider payload into
+    the normalized schema. :meth:`from_file` loads a JSON file and delegates.
+    """
+
+    @classmethod
+    @abstractmethod
+    def from_dict(cls, data: dict[str, Any]) -> AgentTrace:
+        """Parses a raw dictionary input into a canonical AgentTrace."""
+        pass
+
+    @classmethod
+    def _as_float(cls, value: Any, default: float = 0.0) -> float:
+        """Coerces an arbitrary value to float, falling back on malformed input."""
+        try:
+            return float(value)
+        except (TypeError, ValueError):
+            return default
+
+    @classmethod
+    def _as_int(cls, value: Any, default: int = 0) -> int:
+        """Coerces an arbitrary value to int, falling back on malformed input."""
+        try:
+            return int(value)
+        except (TypeError, ValueError):
+            return default
+
+    @classmethod
+    def from_file(cls, filepath: str) -> AgentTrace:
+        """Loads a JSON file and parses it into a canonical AgentTrace."""
+        with open(filepath, encoding="utf-8") as f:
+            data = json.load(f)
+        return cls.from_dict(data)