Workflow row
Trigger: pull_request_target. Secrets: available. Actor: external fork. Agent influence: generated workflow edit. Risk: privileged run from untrusted context.
Resource
Map how AI-generated changes can reach build, release, and deployment.
A risk matrix for engineering and security teams using coding agents, IDE assistants, remediation agents, or CI/CD copilots.
What you receive
Concrete outputs for a specific decision.
- Repo permission map
- Branch protection checklist
- Secrets and workflow trigger table
- PR trust-boundary matrix
- Package and deployment authority notes
Work product
Risk matrix preview
Use the matrix to expose how a suggestion becomes code, a pull request, a workflow run, a package change, or a deployment.
PR trust boundary
Record who authored, reviewed, tested, approved, and merged AI-generated changes.
- Agent output
- Human review
- CI result
- Deployment permission
Package path
Map package additions, lockfile changes, build scripts, post-install hooks, and release automation that an AI system can influence.
01
Matrix dimensions
- Who can request, approve, merge, or deploy AI-generated changes?
- Which workflow triggers run with secrets, write tokens, or release authority?
- Which package, dependency, and generated-code paths can affect production?
- Which artifacts prove what the agent proposed, changed, ran, and approved?
Next step