risk-assessment¶
Multi-agent PR risk assessment orchestrator that coordinates four parallel analyzer sub-agents to evaluate a GitHub pull request across complementary dimensions: security risk and breaking changes, test coverage validation, architecture impact and blast radius, and cross-repository intelligence for affected test repos.
The orchestrator extracts PR metadata via gh CLI, loads context from architecture docs and Jira, launches all four analyzers in parallel using the Agent tool, then aggregates results through a decision engine. This is an advisory-only system -- it provides recommendations but never blocks PRs.
Supports headless mode for CI integration and dry-run mode for testing without publishing. Results are published as PR comments and status checks on GitHub.
Plugin: quality-tooling | User-invocable
Diagram¶
Arguments¶
/risk-assessment <PR_NUMBER> <REPO> [--headless] [--dry-run]
| Argument | Required | Default | Description |
|---|---|---|---|
PR_NUMBER |
- | GitHub pull request number to analyze | |
REPO |
- | Repository identifier (owner/repo format) | |
--headless |
- | Run in headless mode for CI integration | |
--dry-run |
- | Run analysis without publishing results to GitHub |
Usage¶
/risk-assessment 123 opendatahub-io/odh-dashboard
/risk-assessment 456 opendatahub-io/kserve --dry-run