diff --git a/README.md b/README.md index a649771..fae7b8e 100644 --- a/README.md +++ b/README.md @@ -4,6 +4,7 @@ [![PyPI](https://img.shields.io/pypi/v/dprovenancekit)](https://pypi.org/project/dprovenancekit/) [![PyPI Downloads](https://static.pepy.tech/personalized-badge/dprovenancekit?period=total&units=INTERNATIONAL_SYSTEM&left_color=BLACK&right_color=GREEN&left_text=downloads)](https://pepy.tech/projects/dprovenancekit) [![Listed in the official OpenAI Agents SDK docs](https://img.shields.io/badge/OpenAI%20Agents%20SDK-listed%20in%20the%20official%20docs-412991)](https://github.com/openai/openai-agents-python/blob/main/docs/tracing.md#external-tracing-processors-list) +[![Open in Spaces](https://huggingface.co/datasets/huggingface/badges/resolve/main/open-in-hf-spaces-sm.svg)](https://huggingface.co/spaces/dpk890/D.P.K) **Your agent skipped its verification step. The final answer still looked right, so every eval and snapshot test passed. DProvenanceKit — regression testing for AI agents — caught the skipped @@ -20,6 +21,8 @@ has zero third-party dependencies. (plan → search → rank → verify → decide), then catches a later run that drops verification and loops its search tool. It flags the regression as HIGH severity, writes a local SQLite trace, anomaly rules, and a shareable HTML report, then prints copy-paste commands for the gate and viewer. +Or skip the install entirely: the same gate, anomaly rules, diff, and HTML report run in a +[live browser demo on Hugging Face Spaces](https://huggingface.co/spaces/dpk890/D.P.K). > Run → Record → Query → Diff → Detect regressions → Gate in CI @@ -32,7 +35,8 @@ anomaly rules, and a shareable HTML report, then prints copy-paste commands for **Guides:** [Regression testing for AI agents](https://dprovenance.dev/guides/regression-testing-for-ai-agents/) · [A CI gate for LLM agents](https://dprovenance.dev/guides/ci-gate-for-llm-agents/) · [DProvenanceKit vs LangSmith](https://dprovenance.dev/compare/dprovenancekit-vs-langsmith/) · -[OpenAI Agents SDK integration](https://dprovenance.dev/openai-agents/) +[OpenAI Agents SDK integration](https://dprovenance.dev/openai-agents/) · +[Live demo (Hugging Face Space)](https://huggingface.co/spaces/dpk890/D.P.K) **It's not just the library** — it ships the surfaces that make reasoning regressions actionable: diff --git a/docs/crewai/index.html b/docs/crewai/index.html index ac736f6..1f33ed6 100644 --- a/docs/crewai/index.html +++ b/docs/crewai/index.html @@ -192,7 +192,7 @@

What gets recorded

# 04 · gate in ci

Then gate it in CI

-

Once runs are recorded, diff any run against a golden baseline and fail the PR when the reasoning drifts. The server-less gate CLI exits non-zero on regression, and the DProvenanceKit Regression Gate GitHub Action drops it into any workflow — see the full gate story on the main page.

+

Once runs are recorded, diff any run against a golden baseline and fail the PR when the reasoning drifts. The server-less gate CLI exits non-zero on regression, and the DProvenanceKit Regression Gate GitHub Action drops it into any workflow — see the full gate story on the main page. Or try the regression gate live in your browser — an interactive demo of the gate, anomaly rules, diff, and HTML report.

diff --git a/docs/index.html b/docs/index.html index 23d37e0..e7e679b 100644 --- a/docs/index.html +++ b/docs/index.html @@ -234,7 +234,7 @@

Your agent skipped a step.
The answer still looked right.
Listed in the official OpenAI Agents SDK docs as an external tracing processor · integration guide

View on GitHub → - PyPI · GitHub Action + PyPI · GitHub Action · Try it in your browser (HF Space)