nisavid

Use when handling any Git-backed change and safe task completion. Use when asked to implement a change in a repository and commit clean checkpoints; review a branch or repository for bugs, including review-only work; create commits or checkpoints; push and verify a remote branch; reconcile with an exact lease; integrate or discard a branch; classify or clean up a worktree; or close out Git work. If a repository task says "In Codex" or "In Claude Code," apply in either harness, even when mutation or publication is forbidden. Do not use for Git explanations or pasted summaries without repository action. Owns safe task-only commits, publication, completion choices, and provenance-aware cleanup without publishing non-task work.

Use when leading work from Cursor, Claude Code, or Codex that may need delegation across Cursor, Claude Code, Codex, Spark, browser/computer-use agents, or separate worktrees; deciding what stays local versus handed off; or coordinating multi-agent work.

Use when the user explicitly asks to get a GitHub branch or pull request merged, shipped, over the line, or closed out end to end: "get this merged", "drive it to merge", "ship this PR", "merge and close out this branch", or resume a previously stated merge/closeout workflow. Do not use for single-step PR operations, PR-description-only, review-only, status-only, issue-triage, check-only, CI-check-only, comment-only, ready-only, open-PR-only, draft-PR-only, or publish-only requests unless the user also states a merge or closeout goal. Requests to inspect checks, blockers, readiness, comments, or policy without asking to continue toward merge do not trigger this skill. Requests that say to open a PR, leave it draft, or that the PR is not ready to merge do not trigger this skill.

Use when the operator asks to get a branch, change set, draft PR, or pull request ready for review end to end: "get this PR review-ready", "make this ready for review", "prepare this draft PR for review", "mark ready for review", "tighten then publish for review", or "review-ready yeet". Do not use for review-only, status-only, PR-description-only, draft-PR-only, publish-only, merge, ship, or closeout requests unless the operator also states a ready-for-review outcome.

Use when working with Graphite `gt` stacks: creating or tracking stacked branches, navigating or reparenting a stack, restacking, submitting or updating stacked PRs, fixing Graphite metadata, or diagnosing stack ancestry and publication state.

Use when PR review loops, bot review reruns, CodeRabbit or Greptile cycles, unresolved pull request comments, ready-for-review, mark ready, merge readiness, blocked merges, stale review threads, requested reviewers, branch closeout, merge tasks, docs-only PRs, or skill-only PRs affect a pull request.

Use when creating or changing a GitHub PR, including drafts, title/body or draft/ready-state edits, `gh pr create/edit/ready`, Graphite submission, fork-sync or fixup PRs, and requests to yeet, ship, publish, or prepare a PR. Do not use for read-only inspection, comments, checks, threads, or merge-only work with unchanged PR text and state.

Use when returning to an existing open PR that already has review comments or threads and needs merge-focused refresh, conflict or CI repair, review-comment triage, scope trimming, or draft replies. Also use when given a PR number and/or branch for a stale, bloated, neglected, or long-running reviewed PR.

Use when an active Codex, Claude Code, or Cursor session needs explicit Hindsight recall, reflection, mental-model retrieval, or broker status through its controller-owned session bridge.

Use when inspecting, projecting, validating, planning, resuming, or reconciling imports from curated Codex memories, Claude memory files, or portable Markdown or JSONL manifests into the managed Hindsight memory system.

Use when interactively configuring or reviewing managed Hindsight machine archetypes, profiles, providers, credentials, banks, harnesses, models, activation, and prior-memory import choices.

Use when the user asks for thermo-nuclear code quality, a deep maintainability audit, code-judo, file-size, spaghetti, abstraction, boundary, or type-contract review.

Use when the user asks for thermo nuclear, thermonuclear, deep review, or a diff audit for bugs, security, breaking behavior, developer-experience regressions, or feature leaks.

Use when the user asks for thermos, double thermo review, or a combined risk and maintainability branch review.

Use when Arch package changes need a host handoff, rebuild, publish, install, deploy, reinstall, downgrade, local pacman repo refresh, or post-change package verification in this repo.

Use when updating, auditing, or extending this repo's package set, checking the 24-hour dependency freshness sweep, or reconciling upstream, Arch, AUR, CachyOS, or Blackcat Informatics recipe drift.

Use when ROCm, vLLM, AITER, FlashAttention, MIGraphX, quantization, profiling, or inference-optimization references need to become package notes, backlog items, scenarios, or troubleshooting docs in this repo.

Use when validating repo-owned inference engines or model lanes through tracked TOML scenarios after package changes or host repairs.

Use when a task needs an isolated hidden Linux desktop or workspace-owned browser: GUI app QA, web/browser/shopping automation, sandboxed app observation, or stale workspace cleanup. Routes agent-workspace-linux MCP tools on demand. Does NOT apply to host desktop/Chrome control, generic MCP setup, or pure code/file edits.

Use when changing native package metadata or payload, installer-generated launcher behavior, packaged runtime helper behavior, updater service or install behavior, or native package shape in the codex-app-linux repository.

Use when Smith, Wielder, or Outfitter work involves Agent Equipment Config policy layers, effective-config explanation, migration preview, or reviewed local Config writes.

Use when executing Issue Ops advisory workflows for issue review, repair, enrichment, refactor, assignment, duplicate review, selection, session pickup, or issue-set orchestration.

Use when modifying package contents or package-local service/config assets under `packages/*`.

Use when package changes in this repo need a host handoff, rebuild, install, reinstall, or post-change verification.

Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.

Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.

Use when operating a maintained repository fork, validating .agents/fork-ops.toml, assessing fork-ops readiness, inspecting upstream tracks, or migrating fork-related agent materials into Fork Ops.