Metadata-Version: 2.5
Name: proofstep-trajectory
Version: 0.1.0
Summary: Proofstep trajectory policy engine — policy-as-code for tool-using agents
Project-URL: Homepage, https://github.com/IlaKhan17/proofstep
Project-URL: Documentation, https://github.com/IlaKhan17/proofstep/tree/main/docs
Project-URL: Repository, https://github.com/IlaKhan17/proofstep
Project-URL: Issues, https://github.com/IlaKhan17/proofstep/issues
License-Expression: Apache-2.0
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Quality Assurance
Classifier: Topic :: Software Development :: Testing
Classifier: Typing :: Typed
Requires-Python: >=3.11
Requires-Dist: proofstep-types
Requires-Dist: pydantic>=2.9
Requires-Dist: pyyaml>=6.0
Description-Content-Type: text/markdown

# proofstep-trajectory

**Trajectory policy engine** — part of [Proofstep](https://github.com/IlaKhan17/proofstep), the CI gate for AI
agents that knows the difference between a regression and a bad day.

Evaluates what an agent *did* — the ordered sequence of tool calls, their arguments, and the state
they left behind — against policies written as reviewable YAML.

```yaml
rules:
  - id: no-send-before-approval
    kind: forbidden_before
    severity: block
    action: gmail.send
    before: approval_received
```

An agent can produce a flawless email and still have sent it before approval. No output evaluator
can detect that; this is what does.

## Documentation

Full documentation lives in the [repository](https://github.com/IlaKhan17/proofstep/tree/main/docs).

Apache-2.0.
