Crash Lab Python SDK
Python SDK for recording Crash Lab traces and uploading them to a Crash Lab API.
Install
pip install -e sdk
Quick start
from crashlab_sdk import CrashLabClient, crashlab_trace
client = CrashLabClient(
base_url="http://localhost:8000",
api_key="crashlab-dev-key",
project_id="22222222-2222-2222-2222-222222222222",
)
with crashlab_trace("example_task", {"user_id": "u-123"}) as rec:
rec.model_step("plan", {"goal": "create issue"}, output={"ok": True})
rec.tool_call("jira.create_issue(commit)", {"summary": "SDK test"})
rec.tool_result("jira.create_issue(commit)", {"issue_id": "OPS-123"})
trace_id = client.upload_trace(rec.to_trace())
print("trace_id:", trace_id)
API
CrashLabClient
CrashLabClient(base_url, api_key=None, project_id=None, timeout=30)
Methods:
upload_trace(trace: dict) -> strcreate_campaign(name, mutation_set, eval_set, mutation_pack_id=None) -> strstart_run(trace_id, campaign_id, thresholds=None) -> strget_run(run_id) -> dict
TraceBuilder
TraceBuilder outputs backend-compatible traces:
{
"trace_version": "0.1",
"source": "sdk",
"task": {"name": "...", "input": {}},
"events": []
}
Recorder
Use a context manager:
from crashlab_sdk import crashlab_trace
with crashlab_trace("task_name", {"input": "value"}) as rec:
rec.model_step("name", {"foo": "bar"}, output={"ok": True})
rec.tool_call("tool.name", {"arg": 1})
rec.tool_result("tool.name", {"result": "ok"})
rec.error(message="something happened", tool="tool.name")
Decorator
from crashlab_sdk import record_trace
@record_trace(client=client, campaign_id="<campaign_id>")
def run_business_logic(x: int) -> int:
return x * 2
The decorator preserves the function return value and attempts to upload/start runs in the background path without interrupting normal execution.
Example script
See examples/sdk_demo.py.
python sdk/examples/sdk_demo.py
Optional environment variables:
CRASHLAB_API_BASE_URL(defaulthttp://localhost:8000)CRASHLAB_UI_BASE_URL(defaulthttp://localhost:3000)CRASHLAB_API_KEYCRASHLAB_PROJECT_IDCRASHLAB_CAMPAIGN_ID(if provided, script starts a run)
Publish to PyPI (CI)
This repository includes .github/workflows/sdk-publish-pypi.yml to publish the SDK directly to PyPI.
Setup:
- Add
PYPI_API_TOKENin GitHub repository secrets. - Ensure the package version in
sdk/pyproject.tomlis new.
Release:
- Push a tag like
sdk-v0.1.1to trigger publish. - Or run the workflow manually from GitHub Actions.
Metadata
Release files for crashlab-sdk 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| crashlab_sdk-0.1.1.tar.gz | 6.3 kB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| crashlab_sdk-0.1.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 13.0 kB
Release files / crashlab_sdk-0.1.1.tar.gz
| Download URL | crashlab_sdk-0.1.1.tar.gz |
|---|---|
| Size | 6.3 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4fc29c461d6b7d290153daee8975a5be94b9ee6625b8e252f9dfdcad405e49fc
|
|
BLAKE2b-256 checksum How to use checksums |
ae4e7ee00360a43848f0392402b332bda49b2eaa968d7aefabb5d21cd5e05957
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Release files / crashlab_sdk-0.1.1-py3-none-any.whl
| Download URL | crashlab_sdk-0.1.1-py3-none-any.whl |
|---|---|
| Size | 6.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
3fe0e0aa0189e4af0373993df53be7fb15d4e569bc86ecb1d50531027fd4b641
|
|
BLAKE2b-256 checksum How to use checksums |
7cacb5debd03fe4ec91d535d48954705f7b7c05c0b048c767a86fe4af14c27de
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Security for crashlab-sdk 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.