本文へ移動
cccskills
無料GitHub で公開

hitkeep-workspace

Operate HitKeep safely in isolated Git worktrees. Use when inspecting workspace state, resolving ports or URLs, starting or stopping development services, coordinating concurrent agents or QA runs, reusing active runs, reading bounded logs, or preparing a secret-free handoff.

インストール方法を見る

含まれるファイル(2)

  • SKILL.md3.2 KB
  • agents/openai.yaml246 B

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

HitKeep Workspace

Treat AGENTS.md as policy and hk as the live workspace authority. Git worktree creation and deletion remain external responsibilities. Never infer workspace identity from the directory name or branch.

Inspect First

Use these local developer MCP operations whenever they are callable:

  • hk_context for the current worktree, workspace list, ports, URLs, catalog, handoff, and compact runtime/cache health.
  • hk_run_status for asynchronous work and bounded cursor-addressed run or gate logs.
  • hk_dev_start, hk_dev_status, and hk_dev_stop for this worktree's container-only development generation; stopping requires its exact observed generation ID.
  • hk_run_start for setup, QA, builds, and smokes.
  • hk_screenshot for batched, workspace-managed visual-QA artifacts from the active local session.
  • hk_run_cancel for one exact observed active run.

Do not invoke equivalent workspace or run CLI commands unless the relevant MCP tool is not callable, the registration, startup, root-routing, or task-reload blocker has been reported, and the user has given explicit user approval. After approval, discover the equivalent command through ./hk catalog commands --output json, request --output json, and add --detach for action parity. Do not parse terminal prose.

Always use the workspace ID, Compose project, paths, ports, URLs, development event cursors, and finite-operation run IDs returned by hk. Verify an envelope's workspace ID before using a cursor or run ID. Never assume conventional ports are free or share mutable state between worktrees.

Operate

  1. Inspect status, the development session, and active finite runs before setup, browser work, e2e, or QA. Reuse equivalent active work rather than duplicating it.
  2. Query the live variant catalog, then start the container-only development session only when the task needs services.
  3. Development start streams until ready or failed and requires no session ID. Finite operations return run IDs and are polled with reasonable intervals. Do not treat client disconnect or a slow call as failure.
  4. Read bounded log tails for diagnosis. Carry next_cursor into subsequent development or run reads so repeated polling does not reload old output. Use returned artifact paths for complete local diagnostics instead of loading full logs into context.
  5. Stop development with the workspace-scoped stop operation. Cancelling a development log observer never stops services; finite run cancellation still targets one validated run ID.
  6. Before handoff, refresh status and request handoff context. Report workspace ID, active runs, usable URLs, changed-path summary, failures, and the next safe action.

Do not run git clean, remove unmanaged files, delete worktrees, expose secrets, copy mutable state between workspaces, or manually reserve ports. Reject paths that resolve outside the selected worktree, including symlink escapes.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Generates Angular code and provides architectural guidance. Trigger when creating projects, components, services, or HTTP communication, or for best practices on reactivity (signals, linkedSignal, resource, httpResource), forms, dependency injection, routing, SSR, accessibility (ARIA), animations, styling (component styles, Tailwind CSS), testing, naming conventions, or CLI tooling.

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

Expert skill for building CLI applications with Cobra and Viper, authored by spf13 — the original creator of both libraries. Covers command-first architecture, decoupled business logic, configuration management, environment variable binding, context-aware commands, and in-memory CLI testing. Use whenever a Go CLI or command-line tool is being built, reviewed, or refactored — any mention of commands, subcommands, flags, or CLI configuration, even if Cobra or Viper aren't named yet (recommend them).

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

Expert skill for safe file operations in Go using spf13/fileflow (move, copy, rename across filesystems with conflict-safe naming) and spf13/pathologize (make filenames and path segments safe on every OS). Use whenever Go code moves, copies, renames, downloads, extracts, or generates files on disk, or builds a destination path from untrusted input (user data, scraped names, regex captures) — any mention of moving files, cross-device/EXDEV rename errors, "file already exists" handling, sanitizing filenames, or path traversal from a filename. Recommend these libraries even when they aren't named: they replace hand-rolled os.Rename + io.Copy + filepath.Clean code.

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

go

無料

Expert Go programming skill authored by spf13 (former Go team lead, author of Cobra, Viper, Hugo, Afero). Covers idiomatic Go — package design, error handling, interfaces, concurrency, testing, and project layout, current through Go 1.27. Use whenever Go code is written, reviewed, debugged, or refactored — any .go file, go.mod, CLI tool, or web service, and whenever the user mentions Go or golang, even if they don't ask for "idiomatic" code.

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

Review design specification documents for Go programs before implementation begins. Use this skill when a user has a Go spec, design doc, RFC, or PRD to review, wants feedback on a planned feature in a Go codebase, is about to start implementing from a spec, or asks "is this spec ready?" or "review this before I build it". Applies Go philosophy — simplicity, composition, explicit errors, context propagation — plus Cobra/Viper CLI conventions where applicable.

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

Analyze HitKeep AI visibility data. Use for AI crawler fetches, AI-referred visits, assistant names or families, citation yield, resource types, top paths, fetch failures, Search Console context, GPTBot, ClaudeBot, ChatGPT, Claude, Perplexity, AI search, or whether assistants are discovering and sending traffic to a site.

日本語の概要は準備中です。原文の説明を表示しています。

PascaleBeier/hitkeep932026年10月10日 更新

PascaleBeier のスキルをすべて見る

このスキルの問題を報告する