AGENT SKILL

skill-installer

Discover, fetch, validate, and install Grok skills from local directories, GitHub repository URLs (or tree paths), or known curated collections. Creates the proper directory structure under the chosen scope (~/.grok/skills/, <repo>/.grok/skills/, or ./ .grok/skills/), validates the SKILL.md frontmatter, optionally runs a post- install check, and reports exactly what was installed. Use when the user says "install the X skill", "$skill-installer ...", "add this skill from github", "skill-installer", or wants to bring in a ported-from-codex or community skill.

License 许可
MIT
最近更新
2026-07-19
Stars
★ 1
Forks
1

01 前言 Front Matter

name
skill-installer
metadata
{ "source": "ported-from-codex/skill-installer (inferred from Codex $skill-installer behavior + vendor_imports)", "original_codex": "skill-installer (built-in Codex command / $skill-installer)", "short-description": "Install skills from local, GitHub, or curated sources into Grok" }
description
Discover, fetch, validate, and install Grok skills from local directories, GitHub repository URLs (or tree paths), or known curated collections. Creates the proper directory structure under the chosen scope (~/.grok/skills/, <repo>/.grok/skills/, or ./ .grok/skills/), validates the SKILL.md frontmatter, optionally runs a post- install check, and reports exactly what was installed. Use when the user says "install the X skill", "$skill-installer ...", "add this skill from github", "skill-installer", or wants to bring in a ported-from-codex or community skill.
when-to-use
Use when the user wants to install a new or updated skill (local path, GitHub URL, or named curated skill) into one of the Grok skill search locations.
argument-hint
<skill-name-or-path-or-github-url> [to <scope: user|project|local>]

02 资源文件 1

03 SKILL.md 正文

Skill Installer

Install or update Grok skills safely and repeatably.

Scopes (in priority order for the running Grok)

  • Local (CWD): ./.grok/skills/<name>/
  • Project (repo root): <repo-root>/.grok/skills/<name>/
  • User: ~/.grok/skills/<name>/

The skill will ask or infer the best scope (default Project if inside a git repo, else User).

Step 1: Parse the Request

Accept inputs like:

  • my-skill (name only — will search known locations or prompt for source)
  • ./path/to/local/skill-dir
  • https://github.com/org/repo/tree/main/skills/my-skill
  • install the create-plan skill from .experimental
  • Named curated skills (the skill can maintain a small built-in catalog or fetch a manifest)

Resolve the source to a concrete location (local dir or clonable GitHub URL).

Step 2: Fetch / Copy the Skill

  • Local path: Validate it exists and contains SKILL.md. Copy (or symlink in dev mode) the entire directory.
  • GitHub URL: Use grok_com_github MCP tools (get_file_contents on SKILL.md + supporting files, or fall back to run_terminal_cmd git clone --depth 1 --filter=blob:none + sparse checkout of the skill subtree). Or use open_page / web_fetch + manual reconstruction for simple cases.
  • Create the target directory under the chosen scope using run_terminal_cmd mkdir -p or by writing files via the write / search_replace tools.

Step 3: Validate the Skill (Use the Helper Script)

Always run the bundled validator:

python3 "<skill-dir>/scripts/install_skill.py" validate "<path-to-candidate-skill-dir>"

The script checks:

  • Presence and parseability of SKILL.md
  • Valid YAML frontmatter with required name and description
  • name matches the directory name (or warns)
  • No obvious dangerous content (e.g. exec in frontmatter)
  • Presence of referenced scripts/ (they must exist if declared)
  • Basic size / structure sanity

Only proceed on clean validation (or explicit user override after showing warnings).

Step 4: Install / Write the Files

  • For each file in the source skill (SKILL.md, scripts/, references/, templates/*, etc.):
    • Read source content (via read_file for local, MCP get_file_contents or web for remote).
    • Write to the target location using the agent's write tool (or search_replace with empty old_string for new files).
  • Preserve executable bits on scripts where possible (the helper can emit chmod +x commands for the agent to run).
  • Record an install receipt: installed-from.json inside the target skill dir with source, commit-ish (if git), timestamp, and validation result.

Step 5: Post-Install

  • Tell the user the exact final path.
  • Instruct: "Run grok inspect --json or /skills to verify. The skill should appear within seconds (auto-reload on disk change)."
  • Offer to immediately test it: "Would you like me to inject /<name> now and run a quick smoke test?"
  • If the skill declares a post_install hook or test command in metadata, surface it.

Step 6: Update / Re-install

If the target already exists:

  • Show a diff summary (or full git diff style if the target is tracked).
  • Ask for confirmation before overwriting.
  • Support --force / "update in place".

The Helper Script

scripts/install_skill.py (real Python, invoked via run_terminal_cmd):

  • Subcommands: validate, install, diff, receipt
  • Understands both local filesystem paths and can be extended for GitHub tree URLs.
  • Is the single source of truth for "is this a valid Grok skill?"
  • Designed to be run from the workspace root; resolves the target skill dir relative to SKILL.md location of this installer skill when needed.

See the script source for exact CLI and behavior.

Success Criteria

  • The installed skill is immediately discoverable by Grok (appears in /skills and grok inspect).
  • All files from the source are present and bit-for-bit identical (modulo line endings on Windows).
  • Validation passed or user explicitly accepted the warnings.
  • The user knows exactly where it landed and how to invoke it.
  • No silent overwrites of user-modified skills.

Safety & Ethics

  • Never install skills from untrusted sources without explicit user confirmation.
  • Show the full source URL / path and a summary of what will be written before writing anything.
  • If the skill contains references/ or large assets, warn about disk usage.
  • Support "dry-run" mode that only prints the plan.

For This Repo (Prime Gap Structure)

When installing skills here, prefer the project scope (<repo>/.grok/skills/) so the whole team benefits. After install, review the new skill against AGENTS.md before using it on PGS work.

Related Skills

  • create-skill (the bundled interactive creator) — use this installer to bring in ports or community skills; use create-skill for brand-new ones authored in-session.

04 同仓库更多 Skills

zfifteen
zfifteen/skills-catalog

Zotero

Use Zotero Desktop from Codex to enable/probe the local API, search a local Zotero library, list items/collections/tags, export BibTeX, insert citation keys into LaTeX or Markdown drafts, read indexed full text when requested, and import BibTeX/RIS records into Zotero through the connector server. Use when the user mentions Zotero, citations, references.bib, BibTeX export, local Zotero API, localhost:23119, or adding citations from a Zotero library.

★ 1
zfifteen
zfifteen/skills-catalog

zoom-apps-sdk

Use when using Apps SDK.

★ 1
zfifteen
zfifteen/skills-catalog

youtube

Obtain accurate, timestamped YouTube video transcripts (preferring manual captions over auto-generated) as the primary deliverable, then derive summaries, key claims, or notes from that transcript. Use when the user provides a YouTube URL (or video ID) and wants the transcript, a faithful summary, key claims with attribution, timestamped notes, or "what does this video actually say". The skill is transcript-first: never substitute metadata, title, description, or search snippets for the spoken content. Triggers: "youtube transcript", "get the transcript of this video", "summarize this youtube video", "what did they say in https://youtube.com/...", "key claims from this talk".

★ 1
zfifteen
zfifteen/skills-catalog

yeet

Publish local changes to GitHub by confirming scope, committing intentionally, pushing the branch, and opening a draft PR through the GitHub app from this plugin, with `gh` used only as a fallback where connector coverage is insufficient.

★ 1
zfifteen
zfifteen/skills-catalog

x-post-creator

Create X-ready post threads and co-located story packages from the active conversation or referenced project material. Use when the user says turn this into an X-post, /x-post, make this into an X thread, or wants plain-language social posts with supporting images saved to docs/x-posts/[story-name]/.

★ 1
zfifteen
zfifteen/skills-catalog

x-api

Search, read, and write X posts through the local X API MCP server (the same backend used by the Codex x-api skill). Use for searching X/x.com, reading threads and replies, creating lists, staging/publishing posts for @alltheputs, finding high-visibility reply targets for research outreach, ingesting public context, or any request that mentions /x-api or "search X".

★ 1