Metadata-Version: 2.4
Name: agent-trace-card
Version: 0.1.1
Summary: Portable cards for reviewable agent traces and failures.
Author: AuraOne
License-Expression: MIT
Project-URL: Homepage, https://auraone.ai/open
Project-URL: Source, https://github.com/auraoneai/agent-trace-card
Classifier: Development Status :: 3 - Alpha
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# agent-trace-card

`agent-trace-card` generates portable Markdown, HTML, and JSON cards for agent traces. A card summarizes one agent run or failure: goal, outcome, tools used, retries, data touched, policy/rubric checks, failure mode, human intervention, and regression status.

## Quickstart

```bash
python -m venv .venv
. .venv/bin/activate
pip install agent-trace-card
agent-trace-card generate --from examples/refund_trace.json --out card.md
agent-trace-card validate examples/refund_agent_card.json
```

Bundled examples include a failed duplicate-refund trace, a passing read-only trace, and a human-intervened support-review trace. They are synthetic tutorial data only.

## What This Is Not

This is not an observability backend, agent framework, safety certification, or benchmark. It is a shareable review artifact that makes agent traces easier to discuss and turn into regression work.
