skills/skill-select/SKILL.md
Discover and select relevant agent skills based on a problem description, goal, or repository context. Use this to determine which skills apply to a workspace, or when you are unsure which tools are best suited for your current task.
npx skillsauth add ithinkihaveacat/dotfiles skill-selectInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
This skill provides discovery, selection, and management of agent skills for your workspace. It consists of two primary tools:
skill-select: The discovery engine. It analyzes your workspace to suggest sensible default skills, or uses LLM-based selection when task context is provided.skill: The workspace manager. It installs and tracks skills in your workspace, automatically adapting to your environment (Git, Perforce, or Unmanaged).skill-select)Invoke the selection tool via skill-select (symlinked in bin/):
skill-select [DIR] [OPTIONS]
--context TEXT: A comprehensive, self-contained explanation of your task to guide LLM-based selection.--search-dirs PATHS: Colon-separated extra paths to search for skills, overriding SKILL_SOURCE_DIRS.--catalog: Print the full catalog of available skills and exit.--json: Emit structured JSON output instead of formatted text.skill)The skill tool (symlinked in bin/) manages agent skills as untracked symlinks in your workspace. It automatically detects your environment and applies the correct tracking and ignoring mechanism.
skill <command> [arguments]
.agents/skills/ and .claude/skills/. It automatically updates .git/info/exclude to keep your git status clean without dirtying the shared .gitignore.configs/users/<username>/_agents/skills/ and tracks them in .p4-skills-managed in that directory. This is typically used in environments with centralized user configurations..agents/skills/ and tracks them in .p4-skills-managed at the client root..agents/skills/ and .claude/skills/ and tracks them in a local .skills-managed file.apply: Provision recommended skills for this workspace via skill-select (idempotent).suggest: Print recommendations without installing them.add SPEC...: Add a skill (a local path or a registered catalog entry).remove NAME... (alias: rm): Remove a managed skill.list: List skills currently managed in this workspace.update SPEC...: Re-fetch a registered catalog entry (--all for all, --catalog for catalog index).clean: Remove all managed skills and clear tracking records.doctor: Diagnose drift between desired and on-disk skills (read-only).catalog: List all registered skills and their sources.resolve NAME: Print the source path a skill name would resolve to.Analyze the current directory and automatically install the recommended skills:
skill apply
If you have a specific task or hit a roadblock, ask for recommendations based on your situation:
skill suggest --context "Goal: Implement a Wear OS tile in Kotlin. Emulator keeps crashing with OAuth errors."
Add a specific skill from the catalog or a local path:
skill add coding-standards
skill add /path/to/custom-skill
See what is currently active in your workspace:
skill list
Remove all installed skills and restore the workspace to its original state:
skill clean
tools
Provides a comprehensive guide and ADB workflows for testing Android applications (both phone and Wear OS). Focuses on triggering system state changes, simulating connectivity edge cases, implementing robust UI automation, and interacting with Wear-specific surfaces. Triggers: android testing, wear os, testing, adb, pixel watch, galaxy watch, spoofing, fused location.
tools
Discover and select relevant agent skills, and manage workspace tool execution permissions. Use this to determine which skills apply to a workspace, to install or remove skills, and to manage allow/deny/ask rules for local agent tool execution across agents (Claude Code, Antigravity).
documentation
Use this skill when authoring, reviewing, or editing technical documents, including bug reports, known issues, friction logs, PR descriptions, and the structural content and tone of commit messages. Use to ensure engineering content maintains a clear, factual, and constructive tone. Triggers: technical writing, bug report, known issue, friction log, PR description, pull request, commit message tone, review document.
tools
Extracts image URLs and listing metadata from inigo.com property listings. Pairs well with an LLM image-query tool for triaging downloaded photography (e.g. find rooms with a bedside table, a fireplace with art above, mismatched dining chairs). Captures the Inigo-specific JSON paths (the React Server Components chunk format the App Router site uses). Use when scraping inigo.com listings or cataloguing interior-design reference photos. Triggers: inigo, inigo.com, interior design, property photography, gallery scrape, image triage, bedside table, fireplace, mismatched chairs.