dotfiles

Triage this Framework Desktop's Hyprland, CachyOS, SDDM login, Ghostty/Chrome focus, workspace routing, Bluetooth, AirPods, PipeWire, media controls, and local desktop service incidents with live-state checks before changing config.

Triage this Framework Desktop's Hyprland, CachyOS, SDDM login, Ghostty/Chrome focus, workspace routing, Bluetooth, AirPods, PipeWire, media controls, and local desktop service incidents with live-state checks before changing config.

Create GitHub pull requests from the current checkout, including combined commit-and-PR requests. Use when Claude needs to inspect the change, use `draft-commit-message` as the source of truth for the commit message and PR title/body, then push and create a draft or ready PR without a separate wording approval when the user already requested PR creation.

Inspect, instrument, and control Linux Google Chrome through the local Codex Chrome Extension native messaging bridge. Use when Codex needs Chrome tabs, history, navigation, Chrome DevTools Protocol commands, native host manifest/socket troubleshooting, or a Linux browser-control path instead of Chrome DevTools MCP.

Create GitHub pull requests from the current checkout, including combined commit-and-PR requests. Use when Codex needs to inspect the change, use `draft-commit-message` as the source of truth for the commit message and PR title/body, then push and create a draft or ready PR without a separate wording approval when the user already requested PR creation.

Use the local linux-computer-use CLI or MCP server to inspect and safely interact with running Wayland/Hyprland desktop apps through Sky-compatible computer-use tools, grim screenshots, AT-SPI, wtype, ydotool, approvals, and dry-run raw input.

Rename and reorganize saved Codex threads from their recent user prompts and final model responses. Use when Codex needs to review recent thread history, generate concise resume-picker titles, apply thread names through app-server, or run the nightly thread-organizer automation while excluding unattended automation and subagent threads.

Upgrade and audit the local Codex stack across the Arch Linux CLI package, the Linux Codex App repackager, live and dotfiles-managed configuration, system skills and plugins, background services, and current feature maturity. Use when the user asks to upgrade Codex, refresh the latest alpha, update codex-app, audit dotfiles after a release, reconcile runtime or version drift, or produce a report covering stable, experimental, under-development, deprecated, and removed Codex features.

Inspect, list, summarize, or export logged-in ChatGPT.com threads through the local Chrome Bridge.

Maintain the Buck2 BuildBuddy patch stack in a local Buck2 checkout by rebasing fork/stack onto origin/main, publishing bazel-rbe-style merge commits to fork/main with origin/main as first parent, triggering or polling BuildBuddy Workflows, and fixing failed stack commits.

Create, inspect, update, pause, resume, delete, migrate, or troubleshoot Codex automations across backends. Use when Codex needs to manage reminders, recurring jobs, repo monitors, thread follow-ups, local codex-automaton/Aton jobs, systemd user timers, Codex App automations, automation TOML, automation threads, journals, or the localhost Automaton WebUI. On Linux, default to the local codex-automaton Aton/systemd backend; on macOS, default to the Codex App automation backend.

Iteratively finish an active development task by implementing in the main thread, launching clean-room subagent code reviews of the total diff against the target branch, fixing all valid findings, repeating until no findings remain, then committing with draft-commit-message. Use when the user asks Codex to finish a coding task end-to-end with independent review loops before commit.

Create, inspect, or cancel one-time Codex wake-up automations backed by systemd user timers, especially when Codex must re-evaluate time-sensitive state in the future, preserve prior context or approvals, resume a known thread, or install a guarded future action through local codex-automaton.

Use when the user wants to list, inspect, or import ChatGPT-hosted Codex Security Cloud or Aardvark scan findings and accessible repos using local Codex ChatGPT credentials.

Run the Antigravity CLI `agy` as a fast, advisory, read-only scout from Codex. Use when Codex should delegate bounded codebase reconnaissance, first-pass review, hypothesis generation, or evidence extraction to Gemini 3.5 Flash through `agy --print`, while keeping final judgment, edits, and validation in the main Codex thread.

Draft, rewrite, and review Git commit messages and pull-request title/body source text using the Git project's MyFirstContribution guidance. Use when Codex needs to turn staged changes, a diff, commit range, or rough summary into a `component: subject` line, imperative mood, a 50-character-or-less subject, a blank line before the body, and enough reviewer-facing context to explain and sell the change.

Research and implement Buck2 adoption work that bridges Bazel ecosystem expectations, including rules_go or Gazelle comparisons, BCR strategy, Starlark API compatibility, BuildBuddy Workflow bootstrap, release assets, and fixture-backed validation.

Maintain the Buck2 BuildBuddy patch stack in a local Buck2 checkout by rebasing fork/stack onto origin/main, publishing bazel-rbe-style merge commits to fork/main with origin/main as first parent, triggering or polling BuildBuddy Workflows, and fixing failed stack commits.

Reproduce a specific Bazel remote action from a BuildBuddy invocation and generate a customizable `bb execute` replay command. Use when a user provides an invocation URL/ID and wants to re-run one action, modify command args/env/exec properties, or pin execution to a specific executor using scheduler debug properties.

Fetch and triage recent BuildBuddy flaky-test data with BuildBuddyService RPCs. Use when Claude needs to list the latest flaky targets, match the Test Analytics flakes UI, get sample flaky invocations/log pointers, rank recent flakes, check whether a PR is already addressing a flaky target, or start reproducing and fixing the top flaky test.

Compare two BuildBuddy invocation URLs or IDs to troubleshoot hermeticity/reproducibility and cache invalidation. Use when asked to diff canonical Bazel flags, find the first shared action-cache misses, inspect Action/ActionResult differences (inputs/command/env/platform/outputs), or analyze compact execution logs with `bb explain`.

Troubleshoot BuildBuddy invocations via BuildBuddyService (HTTP JSON or gRPC). Use when asked to debug a BuildBuddy invocation, find failed targets and stdout/stderr, inspect remote cache vs RBE metadata, retrieve execution details/profiles/logs, download raw build events or Bazel profiles, or fetch cache scorecard data.

Diagnose BuildBuddy or bb CLI log snippets and customer reports when there is no clean invocation URL, especially FastCDC capability warnings, BES upload failures, confusing invocation IDs, cache proxy behavior, executor UI questions, and source-backed root-cause analysis.

Route public BuildBuddy troubleshooting, invocation comparison, action reproduction, log-root-cause, flaky-test, and usage-analysis requests to the focused BuildBuddy skills bundled in this plugin.

Analyze BuildBuddy billing usage and trends data, fetch GetUsage/GetTrend/GetStatHeatmap/GetStatDrilldown/SearchInvocation/SearchExecution data, compare periods, spot anomalies, and recommend cost-saving actions while protecting API keys.

Design and review landing pages for early-stage B2B devtools, developer infrastructure, AI developer tools, and technical SaaS startups. Use when Claude is asked to create, redesign, critique, or review changes to a startup homepage, launch page, hero section, messaging hierarchy, product-proof section, CTA strategy, visual direction, responsive implementation, or conversion-focused landing page for developers, platform teams, security teams, product engineers, or technical buyers.

Research and answer Bazel questions by reading the Bazel source tree, commit history, and GitHub context (Issues, PRs, Discussions). Use when the user asks how Bazel works, why behavior changed, when a feature/regression appeared, or asks version-specific Bazel questions tied to bazelbuild/bazel.

Create, inspect, update, pause, resume, or delete Claude Code automations for reminders, recurring jobs, repo monitors, and thread follow-ups. Use when Claude needs to turn a natural-language task plus a schedule into a scheduled or recurring automation, avoid creating a duplicate automation, or adjust an existing automation's prompt, cadence, status, or workspace.

Inspect Bazel Central Registry (BCR) modules and bzlmod dependencies. Use when Claude needs to find BCR modules or versions, audit which direct deps in a repo are upgradeable, compare current pins to live BCR releases, update `bazel_dep` or `single_version_override` entries in `MODULE.bazel` (including `include()` files), or inspect module dependency trees.

Resolve Bazel labels from terminal output and open the matching BUILD file target with the local bazel-target-open CLI, especially when tmux-fingers selected labels such as //pkg:target or @repo//pkg:target.

Inspect, instrument, and control a local Google Chrome through the Chrome DevTools MCP server. Use when Claude needs live Chrome tabs, navigation, DOM/console/network inspection, or Chrome DevTools Protocol commands against a running browser, rather than the headless browser automation covered by the playwright skill.

Build a composable CLI for Claude from API docs, an OpenAPI spec, existing curl examples, an SDK, a web app, an admin tool, or a local script. Use when the user wants Claude to create a command-line tool that can run from any repo, expose composable read/write commands, return stable JSON, manage auth, and pair with a companion skill.

Draft, rewrite, and review Git commit messages using the Git project's MyFirstContribution guidance. Use when Claude needs to turn staged changes, a diff, or a rough summary into a commit message with a `component: subject` line, imperative mood, a 50-character-or-less subject, a blank line before the body, and a body that explains the why/context at roughly 72 columns.

Create, repair, validate, preview, and package Codex-compatible animated pet spritesheets from character art, screenshots, generated images, or visual references. Use when a user wants to hatch a Codex pet, create a custom animated pet, or build a built-in pet asset with an 8x9 atlas, transparent unused cells, row-by-row animation prompts, QA contact sheets, preview videos, and pet.json packaging. This skill composes the installed imagegen system skill for visual generation and uses bundled scripts for deterministic spritesheet assembly.

Use the local linux-computer-use CLI or MCP server to inspect and safely interact with visible Hyprland desktop apps through screenshots, AT-SPI state, app approvals, dry-run raw input, and MCP-compatible Computer Use tool calls.

Maintain a friendly fork or long-lived patch stack by refreshing from the origin default branch when requested, rebasing local changes, detecting commits already merged upstream, validating every replayed commit with build/test commands, resolving conflicts without hiding failures, squashing newly added commits into the existing stack, regrouping or renumbering reviewable topic sets, checking tree equivalence for topic-only restacks, and force-pushing rewritten history to the tracked upstream branch with --force-with-lease when asked.

Review recent Claude Code session logs for repeated issues involving dotfiles-managed personal skills, then update or add a personal skill only when the evidence is concrete and recurring. Use for "scan recent sessions for skill issues", "should this become a skill?", or recurring automation runs that maintain personal skills.

Use when the task requires automating a real browser from the terminal (navigation, form filling, snapshots, screenshots, data extraction, UI-flow debugging) via `playwright-cli` or the bundled wrapper script.

Prepare or audit Git worktrees for Claude automation, scheduled/background agents, long-running tasks, or isolated branch work. Use when Claude needs to create a worktree, refresh a worktree, make a worktree ready for unattended execution, or replicate local-only context such as AGENTS.md, user.bazelrc, .bazelrc.user, .buckconfig.local, credentials-backed build config, editor/tooling config, or repo-specific ignored files from a primary checkout.

Use when the user explicitly asks for a desktop or system screenshot (full screen, specific app or window, or a pixel region), or when tool-specific capture capabilities are unavailable and an OS-level capture is needed.