Kheprilab builds tools for working with coding agents: defining the task, reviewing the result, and understanding the code before you take responsibility for it.
You still have to maintain the code.
A passing test does not tell you why a change was made or what will break when you edit it. Wepwawet keeps the task and its acceptance criteria alongside the code, with separate agents for review and, when enabled, checking your understanding.
Define the task
Record the goal, acceptance criteria, and validation commands before implementation.
Review the result
With the judge enabled, a separate agent reviews the change against the task criteria and validation evidence.
Explain the change
Enable comprehension checkpoints to work through the change with an agent that asks questions about the code.
Wepwawet
A harness for coding agents.
The wep CLI installs project context, task templates, and workflow instructions for Codex, Claude Code, GitHub Copilot, and Pi. Task records stay in the repository for later sessions.
# install the harness for your agent $ wep init --target codex # check the setup $ wep doctor
MIT licensed · CLI: wep
About Kheprilab
Independent lab
Kheprilab builds tools for developers working with coding agents, with a focus on code review and understanding.
Open core
The core of Kheprilab’s products is released under the MIT license.
Contact
For questions about the lab or its work, write to hello@kheprilab.com.
Khepri
In ancient Egypt, the scarab who lifts the sun into the sky every morning. The name comes from kheper: to come into being.
Wepwawet
The canid carried on a standard at the front of royal processions. The name means "the one who opens the ways".