dash-docset-install-generate（dash docset 安装生成）

Dash Install and Generate

Overview

Use this skill for install and generation workflows after discovery from $dash-docset-search or direct user request.

Defaults:

• confirmation-first installs
• no silent installation
• hybrid generation: automate known CLI paths, otherwise guide manual Dash generator steps

Use references/dash_url_and_service.md, references/dash_http_api.md, and references/dash_mcp_tools.md during execution.

Automation Prompting

• Codex App automation fit: Guarded. Install and generation flows can require user approval or UI-level interaction.
• Codex CLI automation fit: Guarded. Prefer dry-run or planning mode unless explicit install approval is provided.
• Use references/automation-prompts.md for Codex App and Codex CLI templates with placeholders and stop conditions.
• Default automation behavior should be non-destructive unless install mode is explicitly approved.

Install Workflow

1) Resolve candidate

1. Match request against catalog snapshots:

- uv run python scripts/dash_catalog_match.py --query "<text>"

1. Prefer in this order:

- built-in Dash Downloads - user-contributed - cheatsheet - generation path

2) Perform install (confirmation-first)

Preferred automation path:

• uv run python scripts/dash_url_install.py --repo-name "<repo>" --entry-name "<entry>"
• add --version "<version>" when explicit version is required
• use --yes only when explicit user approval exists

If URL install details are ambiguous:

• switch to guide mode
• provide exact Dash UI path in Settings > Downloads

3) Verify install

Use this order:

1. Dash MCP list_installed_docsets
2. local HTTP /docsets/list (after scripts/dash_api_probe.py)
3. if API unavailable, ask user to confirm from Dash UI

Generation Workflow (Hybrid)

1) Decide automation vs guide

Automate only when there is a documented and stable toolchain for the target source.

Typical examples:

• Sphinx/PyDoctor via docset tooling
• Javadoc pipelines
• GoDoc/Rust/Scaladoc/HexDoc toolchains
• cheat sheet authoring via cheatset

If the path is unclear or brittle:

• guide through Dash Docset Generator and official docs

2) Generation sources to cover

• Swift packages
• Ruby gems
• Python/Sphinx docs
• PHP packages
• Java packages
• Go packages
• Rust crates
• Scala packages
• Dart packages
• Haskell packages
• Hex packages
• Clojure libraries
• generic HTML docs

3) Last-resort generation

Suggest GitHub-based or Stack Overflow-based generation only when built-in, contributed, and known generator paths are exhausted.

API and Service Fallbacks

If MCP is unavailable:

1. run uv run python scripts/dash_api_probe.py
2. use local HTTP API if healthy
3. otherwise provide:

- AppleScript open location "dash://?query=..." - terminal open "dash://?query=..." - macOS Look Up in Dash service guidance

Scripts

• scripts/dash_api_probe.py
• scripts/dash_catalog_refresh.py
• scripts/dash_catalog_match.py
• scripts/dash_url_search.py
• scripts/dash_url_install.py

References

• references/dash_mcp_tools.md
• references/dash_http_api.md
• references/dash_url_and_service.md
• references/automation-prompts.md
• references/catalog_built_in_docsets.json
• references/catalog_user_contrib_docsets.json
• references/catalog_cheatsheets.json

Optional Visual Troubleshooting

If Dash interface behavior needs verification, use the $screenshot skill if it is available in the user's environment.

Interactive Customization Flow

1. Load current effective customization settings first:

• uv run python scripts/customization_config.py effective

1. Ask targeted customization questions:

• Use references/customization-flow.md to drive knob-by-knob questions.
• Confirm desired behavior changes and safety constraints.

1. Map requested changes to implementation files:

• Update SKILL.md, references/*, and any runtime script files listed in references/customization-flow.md.

1. Persist durable customization state:

• Start from customization.template.yaml defaults.
• Apply approved overrides with uv run python scripts/customization_config.py apply --input <yaml-file>.
• Durable path: ~/.config/gaelic-ghost/apple-dev-skills/<skill-name>/customization.yaml.
• Optional override root: APPLE_DEV_SKILLS_CONFIG_HOME.

1. Report resulting effective configuration:

• Re-run uv run python scripts/customization_config.py effective and summarize final active settings.
• If the user asks to remove customization state, run uv run python scripts/customization_config.py reset.

Use references/customization-flow.md for skill-specific knobs, file mapping, guardrails, validation checks, and example requests.