Afelyon – Turn Jira Tickets into GitHub PRs with AI

TL;DR

TL;DR: Afelyon is a self-hosted AI agent that watches your Jira or ClickUp board, reads ticket descriptions, analyzes your codebase, and opens a detailed pull request on GitHub — automatically, without human intervention.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: afelyon.com
• HN launch thread: news.ycombinator.com/item?id=46835389
• License: Proprietary (self-hosted option available — code never leaves your infrastructure)
• Source last checked: 2026-07-26

What Is Afelyon?

Most AI coding tools work from your cursor position or a chat prompt. Afelyon takes a different angle — it starts from the ticket, not the code. You create a Jira or ClickUp ticket with a description and acceptance criteria, and Afelyon autonomously:

1. Clones your repository
2. Analyzes your codebase to understand conventions, patterns, and framework structure
3. Generates a complete code implementation matching the ticket requirements
4. Runs your test suite
5. Opens a pull request with a detailed description of what changed and why

According to the HN launch thread, the team built Afelyon to solve a specific bottleneck: the gap between “ticket created” and “PR ready for review.” They used it internally for months before open-sourcing the platform.

Tech Stack

The self-hosted version ships with Docker and uses:

• TypeScript throughout the stack
• Fastify for the API server
• Next.js for the web dashboard
• Claude AI as the reasoning engine
• PostgreSQL + pgvector for data storage
• BullMQ + Redis for job orchestration

Integrations

Afelyon currently supports:

• Project management: Jira, ClickUp
• Version control: GitHub
• Notifications: Slack

Coming soon: GitLab and Linear integrations.

Pricing

Afelyon offers a tiered pricing model (from the product site):

• Free — self-hosted trial
• $99/month — solo/team plan
• $299/month — professional plan
• $799/month — enterprise plan

A 30-day money-back guarantee applies to all paid plans.

Security

The product site states SOC 2 compliance with end-to-end encryption. Enterprise customers can self-host, ensuring source code never leaves their infrastructure.

Setup Workflow

Prerequisites

• Docker and Docker Compose installed
• A Jira or ClickUp account (API access)
• A GitHub account with repository access
• A Claude AI API key

Step 1: Deploy with Docker Compose

git clone https://github.com/afelyon/afelyon
cd afelyon
docker compose up -d

The Fastify API server, Next.js dashboard, Redis broker, and PostgreSQL instance all start together.

Step 2: Connect Your Integrations

Open the dashboard at http://localhost:3000 and authorize:

1. GitHub — install the Afelyon GitHub App on your repositories
2. Jira or ClickUp — paste your API credentials and project key
3. Claude AI — enter your API key in the settings panel

Step 3: Create a Ticket

In Jira or ClickUp, create a ticket with:

• A clear description of the feature or bug fix
• Acceptance criteria listed out
• Any relevant context (design docs, mockups, prior tickets)

Afelyon monitors your project board and picks up new tickets automatically.

Step 4: Review the Pull Request

Afelyon opens a PR on GitHub with:

• A summary of the code changes
• Reference to the original ticket
• Test results from your CI pipeline
• A checklist of acceptance criteria marked complete or not met

You review, comment, merge — or request changes by adding a comment on the ticket.

Deeper Analysis

Why Ticket-First Matters

Traditional AI coding assistants (Copilot, Cursor, Claude Code) work best when you already know what to type. They assume a developer is in the driver’s seat. Afelyon reverses this — it treats the ticket as the source of truth, which better reflects how most engineering teams actually work.

The gap between ticket creation and first code review is where projects stall. Afelyon collapses that cycle by making the ticket the trigger, not the prompt.

Self-Hosted vs. Cloud

The self-hosted deployment keeps your code on-premises. This is meaningfully different from hosted AI coding tools that send your codebase to third-party APIs. For teams in regulated industries or those with strict IP policies, this may be the deciding factor.

Comparison with Alternatives

• GitHub Copilot: Chat-based, requires a developer actively coding. Afelyon requires no developer involvement after ticket creation.
• Rowboat: Local AI coworker with memory and built-in surfaces. Afelyon is narrower in scope but deeper on the ticket-to-PR automation loop.
• Copilot+ Agents (Microsoft): Requires a Copilot license and Microsoft ecosystem. Afelyon is stack-agnostic.

Practical Evaluation Checklist

• Can it understand complex multi-file changes?
• Does it handle framework-specific conventions (React patterns, Django models, etc.)?
• How does it behave when the codebase lacks sufficient context for a ticket?
• Does the self-hosted version require ongoing maintenance?
• How accurate are the acceptance criteria checkboxes in the PR description?

Security Notes

• SOC 2 compliance claimed — verify with Afelyon sales before enterprise deployment
• Self-hosted option means your code never reaches third-party servers
• BullMQ + Redis stack requires standard job queue hardening (Redis authentication, network segmentation)
• Claude AI API calls send ticket descriptions and code snippets to Anthropic’s API — account for this in your data classification

FAQ

Q: Does Afelyon work with monorepos? A: The HN launch thread does not specify monorepo support. Check the self-hosted docs for current scope.

Q: What happens if the generated code fails tests? A: Afelyon opens the PR regardless and includes the test failure output. You review and iterate by commenting on the Jira ticket.

Q: Can I self-host without paying for Claude API usage? A: The self-hosted version still requires a Claude AI API key. Running Afelyon incurs Claude API costs proportional to ticket volume.

Q: Is the source code open? A: The HN post says “open-sourcing the full platform.” Verify the GitHub org at github.com/afelyon for the current open-source status.

Conclusion

Afelyon targets a specific, real problem: the ticket-to-first-commit lag that clogs engineering backlogs. By making the project management tool the trigger and GitHub PR the output, it fits into existing team workflows without requiring developers to learn a new tool.

The self-hosted Docker deployment is the strongest differentiator for security-conscious teams. If you want AI-assisted code delivery that starts from Jira rather than a chat prompt, Afelyon is worth evaluating.