Skip to main content
This guide installs PeopleBot so your computer can run its local tools. It does not start an AI assistant or make a model call. If commands and folders are unfamiliar, read the terminal lesson first. Use only the command block for your operating system, and run one line at a time. A wheel is the installable Python package (the .whl file). The source archive contains matching source code (the .tar.gz file). A checksum is a fingerprint of a file’s bytes; comparing it helps detect a wrong or damaged download. The provenance file records where the package came from. You do not need to unzip the wheel or source archive to perform this installation.

​
Prerequisites

  • Python 3.11 or newer with pip and virtual-environment support.
  • Git 2.45 or newer for Git-backed operations.
  • Windows for the v0 Instance-admission and execution paths. The Linux commands below cover package installation and CLI help only.

​
Download and verify

Download all four assets from the 0.1.0a4 release into one directory:
  • peoplebot-0.1.0a4-py3-none-any.whl
  • peoplebot-0.1.0a4.tar.gz
  • peoplebot-0.1.0a4-public-provenance.json
  • SHA256SUMS
Open a terminal in that directory. Verify the checksum file itself against the SHA-256 printed in the GitHub release body. On Windows PowerShell:
Compare the first hash with the checksum printed in the linked GitHub release body, then compare each remaining hash with its filename’s entry in SHA256SUMS. Letter case does not matter. Continue only if all four match. On Linux:
The first hash must match the checksum printed in the linked GitHub release body, and the second command must report OK for all three entries.

​
Install in a virtual environment

On Windows PowerShell, using an installed Python 3.11-or-newer interpreter:
On Linux, using an installed Python 3.11-or-newer interpreter:
No shell activation is needed because these commands use the virtual environment’s Python directly.

​
Expected result

The help output lists the State/alpha commands plus finite work/development and usage-reporting commands. Their presence does not mean an agent or schedule is configured or active. Keep the matching source archive and provenance with the installed wheel. Read release capabilities before attempting an execution. If Python, Git, or virtual-environment creation is unavailable, install the missing prerequisite before continuing. If the package cannot be imported, confirm you are running the virtual environment’s Python shown above. Source: public adoption instructions and CLI implementation.

​
Continue with a real local operation

Resolve your first State using the public source and your installed package. It produces inspectable output without a model account. For command failures, use troubleshooting.