Skip to content

Latest commit

 

History

History
148 lines (110 loc) · 4.3 KB

File metadata and controls

148 lines (110 loc) · 4.3 KB

Installing the Suite

The suite follows the open Agent Skills standard: every skill is a directory with a SKILL.md entry point plus optional references/, templates/, scripts/, examples/, and agents/. Installing for a host means copying the skill directory into that host's skills location, then restarting (or starting a new session).

No host-specific wrapper is required: every skill already ships the name and description frontmatter fields that all SKILL.md-based hosts read.

One copy for most hosts

.agents/skills/ is the vendor-neutral location recognized by opencode, Codex CLI, Trae, and most other SKILL.md-based agents. A single copy there covers them:

mkdir -p .agents/skills
cp -r python-*/ .agents/skills/

Use this for shared projects. Use the per-host paths below for personal, cross-repo installs or for hosts that do not read .agents/skills/.

Per-host locations

Host Global Project
opencode ~/.config/opencode/skills/ .opencode/skills/
Claude Code ~/.claude/skills/ .claude/skills/
Codex CLI ~/.codex/skills/ .codex/skills/
Kilo Code ~/.kilo/skills/ .kilo/skills/
Trae ~/.trae/skills/ .trae/skills/

opencode

# Global (all projects)
mkdir -p ~/.config/opencode/skills
cp -r python-*/ ~/.config/opencode/skills/

# Project
mkdir -p .opencode/skills
cp -r python-*/ .opencode/skills/

opencode also discovers .claude/skills/ and .agents/skills/ in the repo and home directory. Optionally restrict skills with permission.skill in opencode.json (for example "experimental-*": "ask").

Claude Code

# Personal
mkdir -p ~/.claude/skills
cp -r python-*/ ~/.claude/skills/

# Project
mkdir -p .claude/skills
cp -r python-*/ .claude/skills/

The directory name must match the name frontmatter field and be lowercase kebab-case; the suite already complies. Verify with claude --list-skills or /skills inside a session.

The plugin route is supported too: this repository ships .claude-plugin/marketplace.json and .claude-plugin/plugin.json, so

/plugin marketplace add bzdvdn/python-developer-ai-skill
/plugin install python-developer-ai-skill@python-developer-ai-skill

loads all twelve skills as a plugin (the plugin manifest lists each python-<name> directory as a skill). cp -r to .claude/skills/ remains the fallback for hosts or versions without the plugin loader.

Codex CLI

# User (respects $CODEX_HOME)
mkdir -p ~/.codex/skills
cp -r python-*/ ~/.codex/skills/

# Project
mkdir -p .codex/skills
cp -r python-*/ .codex/skills/

The repo-root AGENTS.md also serves as Codex CLI project context; a user-level global file can go in ~/.codex/AGENTS.md.

The plugin route is supported too: this repository ships .codex-plugin/plugin.json, so

codex plugin marketplace add bzdvdn/python-developer-ai-skill
codex plugin add python-developer-ai-skill@python-developer-ai-skill

loads all twelve skills as a plugin. cp -r to .codex/skills/ remains the fallback for CLI versions without the plugin loader.

Kilo Code

# Global
mkdir -p ~/.kilo/skills
cp -r python-*/ ~/.kilo/skills/

# Project
mkdir -p .kilo/skills
cp -r python-*/ .kilo/skills/

Windows global path: %USERPROFILE%\.kilo\skills\. Kilo Code also reads compatible directories such as .claude/skills/ and .agents/skills/. Custom search paths can be declared via skills.paths / skills.urls in kilo.jsonc.

Trae

# Global (macOS/Linux)
mkdir -p ~/.trae/skills
cp -r python-*/ ~/.trae/skills/

# Project
mkdir -p .trae/skills
cp -r python-*/ .trae/skills/

Windows global path: %userprofile%\.trae\skills\. Trae also reads .agents/skills/ once enabled under Settings > Skills & Commands > Enable .agents Skills Directory; .trae/skills/ takes priority on name collisions.

Verification

  1. Restart the host (or start a new session) — skills are scanned at session start.
  2. Confirm SKILL.md sits directly inside each <name>/ directory and starts with name + description frontmatter.
  3. Run the suite's own consistency check from the repository:
python3 scripts/validate_suite.py --strict
  1. Load a skill and check the agent sees it (for example ask opencode which skills it has, or open /skills in Claude Code).