Fetch and prioritize the open llvm/llvm-project PRs awaiting my review. Combines a deterministic GitHub fetch (review-requested / reviewed-by / mentions, deduped and scored by area + signals) with an AI-assisted ranking that surfaces what actually needs my eyes first. Load when asked to triage my review queue, show what PRs to review, or "what should I review today".
Multi-agent, read-only code review of a GitHub PR, a branch, or the current working-tree diff. Runs a fast high-signal pass (correctness, adversarial, design/API) and renders an interim report early so you can decide whether the change is worth more effort, then a deeper pass (tests/docs/conventions plus conditional security/performance/concurrency), gates findings by confidence, validates them in a batched independent pass, and renders a ranked final report. Caps parallel agents hard to stay fast and avoid rate limits. For C++ in an LLVM/Clang/LLDB tree it leans on the llvm-development skill so reviewers apply LLVM conventions and run clang-format/clang-tidy. Never edits, commits, or pushes. Load when asked to review a PR, review changes before opening a PR, or get feedback on a diff.
Conventions for writing, refactoring, or reviewing C++ in LLVM, Clang, or LLDB trees. Covers the naming split between LLVM and LLDB, include ordering, ADT containers and string types (StringRef, SmallVector, DenseMap), Error/Expected handling, LLVM-style casts, formatting rules, and LLDB-specific hazards. Load when editing `.cpp`/`.h` files under an LLVM monorepo, opening a file that includes `llvm/ADT/`, or answering questions about LLVM/LLDB coding standards.
Conventions for writing source code comments. Load when adding or editing comments in code, reviewing a diff for comment quality, or deciding whether a comment belongs at all. Covers when a comment earns its place, the WHY-not-WHAT rule, lifting rationale to general principles rather than concrete examples, banning prompt/task references, and the tone and length to aim for.
Conventions for drafting git commit messages. Load when asked to write a commit message, draft a commit, or run `git commit`. Covers subject-line bracket prefixes, NFC tagging, line-length and wrapping rules, imperative mood, body content focused on the why, and trailers like rdar:// references and `Assisted-by: Claude`.
Guidance for offloading work to subagents and running independent work in parallel. Load when a task involves multiple independent investigations, heavy codebase exploration, or research that would bloat the main context. Covers when to delegate vs. do directly, how to dispatch agents concurrently in a single message, how to choose between Explore / Plan / general-purpose / specialist agents, and how to structure agent prompts for useful results.