I've had some success asking the Antigravity agent to find flaky tests and propose fixes for them, then just running it in the background and reviewing what it finds. Upstreaming this to the repo so others can use it, since it includes helpful notes like `--runs_per_test` and leveraging random seeds as well as an iteration loop.
I opted not to have the agent do anything with PRs just yet, but if this is useful and we build confidence in it, we can explore that in the future.
The current PR review skill blindly attempts to checkout branches using the `gh` CLI.
This creates fatal errors if the branch being checked out is already tied to a different git worktree.
Additionally, the logic didn't give the user a clear way to enforce a remote code review when they are the author.
This change updates the `Execution Workflow` to prioritize an explicit user instruction to review remotely, and updates the `Local Code Review` instructions to enforce checking for worktree conflicts before executing local checkouts.
This commit introduces an AI agent skill for reviewing pull requests
against the Angular repository. It establishes guidelines for ensuring
code cleanliness, performance, testing, API design, and payload size.
It supports performing reviews through both the GitHub CLI (remote) and
local editing, factoring in package-specific guidelines (e.g., router)
and prioritizing user approval before posting comments. By checking existing
comments first, the agent can avoid duplicate reviews.
This provides the AI agent with a reproducible workflow for providing
constructive, manual PR feedback.
Antigravity does not support AGENTS.md, instead it has "workspace rules" (antigravity.google/docs/rules-workflows). This commit adds a symlink for the workspace rule.
Andrew Scott