GitHub

Overview

AI has moved into the software lifecycle - coding assistants, review bots, and agents that run in CI. DataStrict governs the model calls those tools make, so source code and secrets do not leak into prompts and vendor models.

Policy lives as code in your repository and is reviewed through the same pull requests as everything else.

How it works

Model calls from GitHub Actions, agents, or assistant backends are routed through the DataStrict gateway. It inspects each prompt for source, secrets, and tokens, applies policy, and records the decision. Because policy is a file in your repo, changes are diffed, reviewed, and versioned.

Connect

Set the model base URL in your Actions workflow or agent to the DataStrict gateway, and keep your policy directory in the repo so it ships and reviews with your code.

# .github/workflows/ai.yml
env:
  OPENAI_BASE_URL: https://datastrict.internal/v1
# policy-as-code lives in ./policies and is reviewed in PRs

What you can enforce

• Model calls from CI and coding agents evaluated against policy before they run.
• Source, secrets, and tokens blocked from leaving in a prompt.
• Policy diffed and reviewed as code in your repository.
• Every decision recorded to a hash-chained Ledger.