agent-skills

Multi-agent review-and-improve loop for a GitHub PR you have checked out — posts a "starting" PR comment cc'ing the original author, runs requested rounds plus any adaptive continuation, applies fix commits to the local branch after each round, pushes everything back to the PR, then edits the starting comment in-place with the synthesized report (or a failure summary). Auto-detects the PR from the currently checked-out branch when no locator is supplied. Use when the user wants to "improve a PR", "review and commit fixes", "iterate on my PR", or "review and push back" against a checked-out PR branch. Requires `gh`, `uuidgen`, `jq`, and `uv` or `python3` on PATH. Activates the `review-anvil` engine in per_fix mode.

Read-only multi-agent review of a GitHub Pull Request, with the synthesized report posted back as a PR comment so the author is notified. Use when the user wants to review a GitHub PR (github.com or GitHub Enterprise) and post a structured review back to the PR conversation. Auto-detects the PR from the currently checked-out branch when no locator is supplied. Requires `gh`, `uuidgen`, `jq`, and `uv` or `python3` on PATH. Activates the `review-anvil` engine in read-only mode and orchestrates the shell helper for posting.

Multi-agent code review without any edits or commits — requested rounds of parallel reviewers, synthesis, default-on reproduction of uncertain material findings, and a final report. Use when the user wants a code review of a diff, branch, path, or PR but does NOT want any fixes applied automatically (no edits, no commits, no staging). Activates the `review-anvil` engine in read-only mode.

Iteratively refine code via requested rounds of parallel subagent review and orchestrator-applied fixes, with bounded adaptive continuation enabled by default for productive runs. Use when the user says "let's do three rounds", "fix/review loop", "back-and-forth review", "iterative review", or asks to harden a change with multiple rounds of codex/claude review.

Convert academic PDFs to clean Markdown. Use whenever you encounter a PDF paper (local file or URL) that needs to be read, analyzed, or referenced as text.

Delegate code review, plan review, and exploration to Codex CLI. Use when you need an independent second opinion on code changes, plans, or architecture — or when asked to run codex review/exec.

Delegate code review, plan review, and exploration to Claude Code CLI. Use when you need an independent second opinion on code changes, plans, or architecture — or when asked to run claude review/exec.

Recompile an Overleaf project and extract its build logs, errors and warnings in a standardized format. Use when asked to compile/recompile an Overleaf paper, debug why a PDF build fails, see what's broken after pushing changes, or pull LaTeX/biber errors and warnings out of an Overleaf project. Drives an already-logged-in browser tab via the chrome-cdp skill.

Bulk-post Overleaf comments from a JSON plan. Use when asked to add many Overleaf comments at once (e.g. thesis review, paper feedback, multi-line annotations). Drives an existing logged-in browser tab via the chrome-cdp skill.

Bulk-post anchored comments (with @-mentions) to a Google Doc from a JSON plan. Use when adding many Google Docs comments at once — checklist audits, doc reviews, structured feedback — that would otherwise mean clicking through the UI for each one. Drives an existing logged-in browser tab via the chrome-cdp skill.

Upload local images to a GitHub PR description or comment without committing them to the repo. Use when creating PRs that need before/after screenshots, visual diffs, or any embedded images.

Find academic papers, explore citations, and export BibTeX. Use when asked about research papers, references, citations, bibliography, or scholarly information.

Take screenshots of windows on Wayland/GNOME. Use when asked to capture, screenshot, or photograph a window, app, or screen. Supports selection by app name, PID, title, or window ID.