Blog
From the Codepylon team
Engineering deep-dives, product updates, and AI research from the team building autonomous pull request agents in Los Angeles.
AI Research, Engineering
Why context windows alone cannot understand your codebase
Engineering, Product
Teaching an agent to respond to code review comments
Engineering, Case Study
Migrating 80,000 lines of Python 2 with Pylon-Migrate
Product, Security
Role-based access control for autonomous agents: what we built and why
Engineering, AI Research
What makes a generated test actually useful
Product
Closing the loop: Slack notifications that actually tell you something
Engineering
How Codepylon scopes agent jobs in a monorepo without touching the wrong service
Engineering, AI Research
Finding dead code that is actually safe to delete
Product
Codepylon is now open for early access
Product
GitLab support is here
Engineering, AI Research
Why we built a PR-opening agent instead of another autocomplete
Product, Engineering
Closing the ticket automatically: how Pylon links Jira and Linear issues to merged PRs
Engineering
Refactoring incrementally: why big-bang rewrites fail and PRs succeed
Engineering, AI Research
The repository context graph: how Pylon understands a codebase
Engineering
Enabling TypeScript strict mode on a 200,000-line codebase, one file at a time
Engineering, Product
When the agent fails: how Codepylon handles jobs it cannot complete
Product, Engineering
What a good agent-written PR description looks like
Product
Installing the Codepylon GitHub App: what permissions we request and why
Engineering
The first PR Pylon ever opened on a real codebase
Engineering, AI Research