## Problem Statement
The project faces a growing operational challenge: the small core development team manages client-facing work via Jira, while a global community of contributors submits issues and pull requests through GitHub. This creates two gaps:
1. **Triage gap:** Community GitHub issues and core-team Jira tickets are loosely connected. Duplicates, related work, and priority alignment between the two systems require significant manual effort.
2. **Review gap:** The volume of incoming PRs — across the main repo and 8+ submodules (plugins, analyzer bridge, HAPI FHIR starter, data export, Liquibase tools, etc.) — exceeds the core team’s capacity to review thoroughly. PRs can sit unreviewed, or get merged without sufficient quality validation against the project’s constitutional principles.
## Proposed Work
Build an MVP that takes the existing local AI commands and project knowledge, and scales them into an automated, CI-integrated pipeline. The project has three main work streams:
1. **Goal:** When a PR is opened or updated, automatically run project-specific quality checks and post structured feedback — without a core team member having to manually invoke anything.
Concrete work:
- Create GitHub Actions workflows that trigger on PR events (`opened`, `synchronize`, `ready_for_review`)
- Adapt the existing `/audit-branch` command’s detection rules (Tier 1 mechanical + Tier 2 heuristic) into CI-runnable checks that post inline PR comments
- Implement **hard-stop checks** (blocking merge on CRITICAL findings like merge conflicts, i18n duplicate keys, constitutional architecture violations like `@Transactional` on controllers) and **soft advisory checks** (code quality suggestions posted as non-blocking review comments)
- Explore integrating with GitHub’s native **Copilot Code Review** (now running on an agentic architecture as of March 2026, with static analysis integration supporting CodeQL and ESLint) alongside project-specific custom checks
- Evaluate tools like **CodeRabbit** for deeper context-aware review, potentially running alongside Copilot for complementary coverage
### 2. Cross-Platform Issue Triage & Linking
**Goal:** Help bridge the gap between community GitHub issues and core-team Jira tickets, reducing manual triage overhead.
Concrete work:
- Build an AI-assisted triage workflow (GitHub Action or bot) that classifies new GitHub issues by category (bug, feature request, question, duplicate) and priority
- Explore automated linking between GitHub issues and related Jira tickets using the Jira and GitHub APIs
- Investigate leveraging **GitHub’s Agentic Workflows** (in technical preview as of February 2026, supporting plain-Markdown workflow definitions for AI-driven repository tasks) for intelligent issue routing
- Generate periodic triage reports summarizing open issues, stale PRs, and alignment gaps between GitHub and Jira backlogs
### 3. CI Integration & Developer Feedback Loop
**Goal:** Make the automated checks useful and non-annoying — contributors should get fast, actionable feedback that helps them improve their PRs.
Concrete work:
- Design the feedback format: structured PR comments with severity levels, specific file:line references, and suggested fixes (following the pattern established by `/audit-branch`'s report format)
- Implement configurable severity thresholds per check (repo-level configuration, not hardcoded)
- Create a feedback dashboard or summary comment on each PR showing check status
- Ensure the pipeline works across the main repo and key submodules (especially `openelisglobal-plugins` and `openelis-analyzer-bridge`)