Ironic Backport Agent
Automates backporting upstream Ironic changes across OpenShift release branches using Claude AI for intelligent decision-making and natural language interaction.
README
Ironic Backport Agent (MCP Version)
AI-assisted automation for backporting upstream Ironic changes across OpenShift release branches.
Works with Claude Code - No separate API keys needed!
Project Goal
Automate the repetitive backport workflow for OpenShift BareMetal/Ironic releases using Claude AI for intelligent decision-making. Runs as an MCP server that integrates seamlessly with Claude Code.
JIRA Epic: METAL-1814
Research Spike: METAL-1817
How It Works
Instead of using CLI commands, you talk to Claude naturally:
You: "Backport PR #370 to release-4.20"
Claude: [analyzes PR] "I found 3 commits. Based on the changes,
only commit 42ac11d31 should be backported. Proceed?"
You: "Yes"
Claude: [executes backport, creates PR] "✅ Done! PR #456 created"
Features
🤖 AI-Powered (via Claude Code)
- Analyzes PRs to recommend which commits to backport
- Determines which release branches need the fix
- Suggests merge conflict resolutions
- Generates PR titles and descriptions following OpenShift conventions
- Adapts to errors and retries intelligently
⚙️ Automated Operations
- Git cherry-pick with conflict detection
- Branch management and analysis
- GitHub PR creation with proper formatting
- Dependency file updates (requirements.cachito, main-package-list.ocp)
💬 Natural Conversation
- No commands to memorize - just describe what you want
- Context-aware - Claude remembers the conversation
- Human oversight - You approve critical steps naturally
- Audit trail built into Claude conversation history
Quick Start
1. Install
cd /Users/jhajyahy/ironic-backport-agent
./install.sh
2. Configure
# Only need GitHub token - no Claude API key!
export GITHUB_TOKEN="ghp_xxxxxxxxxxxx"
Get token from: https://github.com/settings/tokens (scopes: repo + workflow)
3. Set Up MCP
Add to ~/.claude/config.json:
{
"mcpServers": {
"ironic-backport": {
"command": "python",
"args": ["/Users/jhajyahy/ironic-backport-agent/src/backport_agent/mcp_server.py"],
"env": {"GITHUB_TOKEN": "${GITHUB_TOKEN}"}
}
}
}
4. Restart Claude Code
The backport tools will now be available in your Claude Code session!
📖 See SETUP-MCP.md for detailed setup guide.
Config File
Create config.yaml:
github:
base_org: openshift
repos:
- openstack-ironic
- ironic-image
- ironic-python-agent-image
claude:
model: claude-sonnet-4-5
max_tokens: 4096
release_branches:
- release-4.17
- release-4.18
- release-4.19
- release-4.20
- release-4.21
- release-4.22
workflow:
auto_approve: false
conflict_strategy: ai_suggest
Usage Examples
Simple Backport
You: "Backport PR #370 from openstack-ironic to release-4.20"
Claude: "I found PR #370 with 3 commits. Analyzing..."
[uses analyze_pr tool]
"Based on the changes:
- 42ac11d31: Update heartbeat ✓ BACKPORT
- abc123de4: Update dev deps ✗ SKIP
- def456g78: Fix pep8 ✗ SKIP
Proceed with backporting commit 42ac11d31?"
You: "Yes"
Claude: [creates branch, cherry-picks, creates PR]
"✅ PR created: https://github.com/.../pull/456"
Check Branch Coverage
You: "Does release-4.20 already have commit 42ac11d31?"
Claude: [uses analyze_branch_coverage tool]
"No, it's not in release-4.20.
Present in: release-4.21, release-4.22
Would you like to backport it?"
Batch Backport
You: "Backport PR #370 to all active releases"
Claude: "Analyzing which branches need this fix..."
[checks each branch]
"Results:
- release-4.18: Needs it ✓
- release-4.19: Needs it ✓
- release-4.20: Needs it ✓
- release-4.21: Already has it ✗
- release-4.22: Already has it ✗
Create 3 backport PRs?"
You: "Yes"
Claude: [creates PRs for 4.18, 4.19, 4.20]
"✅ All done!
- 4.18: PR #457
- 4.19: PR #458
- 4.20: PR #459"
Conflict Handling
You: "Backport commit 42ac11d31 to release-4.20"
Claude: [attempts cherry-pick]
"⚠️ Conflict in ironic/conductor/manager.py
Analyzing..."
[reads conflict content]
"The conflict is in import order. Suggested resolution:
- Use THEIR version for update_heartbeat function
- Keep OURS version for imports
Apply this?"
You: "Yes"
Claude: [resolves conflict, completes cherry-pick]
"✅ Resolved and cherry-picked!"
Architecture
See ARCHITECTURE.md for detailed design documentation.
Development
Run Tests
pytest tests/ -v --cov=backport_agent
Code Quality
# Format
black src/
# Lint
ruff check src/
# Type check
mypy src/
METAL-1817 Research Findings
This prototype addresses all research questions from the spike:
✅ Branch Analysis
- Implemented via
git branch --containsand AI-assisted applicability detection - See:
src/backport_agent/git_ops.py:analyze_branch_coverage()
✅ Cherry-Pick Automation
- Conflict detection with AI-suggested resolutions
- See:
src/backport_agent/git_ops.py:cherry_pick_with_ai()
✅ GitHub Integration
- PAT scope requirements documented above
- PR creation via PyGithub with proper title format
- See:
src/backport_agent/github_client.py
✅ Dependency File Updates
- AI determines which file to update (requirements.cachito vs main-package-list.ocp)
- Automated hash updates
- See:
src/backport_agent/dependency_updater.py
✅ GitLab Integration
- Python-gitlab for ospinfo repos
- See:
src/backport_agent/gitlab_client.py
Roadmap
- [x] METAL-1817: Git/GitHub/GitLab integration research
- [ ] METAL-1820: Phase 1 prototype (Scenario A)
- [ ] METAL-1821: JIRA API integration
- [ ] METAL-1819: CI monitoring
- [ ] METAL-1822: Brew/Koji support (Scenario B)
- [ ] METAL-1823: Optimized semi-autonomous mode
- [ ] METAL-1824: Production deployment
Contributing
This is a team effort! See the Slack channel for coordination.
License
Internal Red Hat tooling - not for public distribution.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.