mokra 0.1.0

See what your AI agent actually did. One line. Zero config.

mokra

See what your AI agent actually did. One line. Zero config.

Installation

pip install mokra

Quick Start

from mokra import observe

with observe() as world:
    # Run your AI agent here
    agent.run("Send a welcome email to new users")

# See exactly what happened
print(world)

Features

• Zero config: Works out of the box with any HTTP-based service
• Deep adapters: Rich insights for Stripe, Slack, SendGrid, and more
• Framework support: First-class support for LangChain and CrewAI
• Test generation: Automatically generate pytest tests from observed behavior

Supported Services

Service	Adapter Type	Features
Stripe	Deep	Payment intents, customers, subscriptions
Slack	Deep	Messages, channels, reactions
SendGrid	Deep	Emails, templates, recipients
Any HTTP	Generic	Request/response logging

Documentation

• Testing LangChain Agents in 60 Seconds
• What Did My Agent Do?

Releases & Versioning

mokra follows Semantic Versioning with automatic version detection from git tags:

MAJOR.MINOR.PATCH

v0.1.0  →  First release
v0.1.1  →  Bug fix
v0.2.0  →  New feature
v1.0.0  →  Major/breaking change

Pre-release versions:

• v0.1.0a1 - Alpha releases
• v0.1.0b1 - Beta releases
• v0.1.0rc1 - Release candidates

Publishing a new version:

git tag v0.2.0
git push origin v0.2.0

That's it! The version is read automatically from the git tag - no need to edit pyproject.toml.

Note: PyPI requires unique version numbers - you cannot overwrite an existing version.

Contributing

See CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE for details.