A public log of my open-source contributions and reviews.
This repository collects two kinds of artifacts:
- Drafted PR reviews — careful, long-form reviews of open pull requests in public OSS projects I follow. Each review walks through the diff, surfaces risks, asks the questions a thoughtful maintainer would ask, and records what I learned from reading it.
- Cross-cutting insights — periodic syntheses (
reviews/INSIGHTS.md) that extract themes across many reviews — what's actually changing in the AI coding-agent ecosystem week to week.
Drafted reviews in this repo are NOT automatically posted to upstream
projects. They are public reading material — a study log. If I decide
to engage upstream on a particular PR, I do so manually, with a focused
comment, after re-reading and trimming. Treat everything under
reviews/ as my notes, not as a maintainer's verdict on the PR.
reviews/
<repo-flat>/ # e.g. anomalyco-opencode/, BerriAI-litellm/
PR-<number>.md
INDEX.md # all reviews, organized by repo
INSIGHTS.md # cross-PR theme extraction
Each review file uses YAML frontmatter and a fixed seven-section structure (Context, Diff walkthrough, Risks, Questions, Verdict, What I learned).
The current rotation of repos I sample PRs from:
- anomalyco/opencode — terminal coding agent
- BerriAI/litellm — model gateway / proxy
- modelcontextprotocol/servers — reference MCP servers
- openai/codex — agentic CLI
- OpenHands/OpenHands — autonomous coding agents
- cline/cline — VS Code coding agent
- Aider-AI/aider — pair-programming CLI
- charmbracelet/crush — terminal coding agent
- Code snippets and scripts: MIT (see
LICENSE). - Prose (review write-ups, insights, synthesis): CC-BY-4.0.
Quoted diff snippets remain under the upstream project's license; they are included here under fair-use review/commentary, with attribution to the source PR via the URL in each file's frontmatter.
See reviews/INDEX.md for the full list, and
reviews/INSIGHTS.md for cross-PR themes.