TypingMindTypingMindThe best frontend for AI
Skilldeck logo

Skilldeck

Community
crossoverJie
skilldeck

Manage AI agent skills via natural language — install, update, delete, and assign skills to agents using SkillDeck's filesystem conventions.

Overview

PublishercrossoverJie
RepositorySkillDeck
Skill nameskilldeck
Stars
572
Forks
40
Bundled files
Instructions only
LicenseMIT
Links
Skill sourceRepositoryHomepage
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by crossoverJie on GitHub. Read the source before you install it.

Installation

Install the Skilldeck AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/crossoverJie/SkillDeck/tree/main/skills/skilldeck
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/crossoverJie/SkillDeck.git /tmp/SkillDeck
mkdir -p .claude/skills
cp -r /tmp/SkillDeck/skills/skilldeck .claude/skills/skilldeck
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Skilldeck in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Skilldeck on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Skilldeck is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

SkillDeck Manager

Manage AI coding agent skills through natural language. This skill teaches you how to install, update, delete, and assign skills using SkillDeck's filesystem conventions.

SkillDeck is a macOS GUI app that manages skills across 14 AI agents. Skills are stored as directories containing SKILL.md files. This skill lets you perform the same operations from the terminal.

Filesystem Layout

~/.agents/
├── skills/                    # Canonical storage (real files live here)
│   ├── my-skill/
│   │   ├── SKILL.md           # Skill definition (YAML frontmatter + markdown)
│   │   └── ...                # Other files (scripts, templates, etc.)
│   └── another-skill/
│       └── SKILL.md
├── .skill-lock.json           # Lock file (v3, shared with npx skills)
└── .skilldeck-cache.json      # SkillDeck's private cache (do not modify)

~/.claude/skills/              # Claude Code skills (symlinks → canonical)
~/.codex/skills/               # Codex skills (symlinks → canonical, also reads ~/.agents/skills/)
~/.gemini/skills/              # Gemini CLI skills
~/.copilot/skills/             # Copilot CLI skills (also reads ~/.claude/skills/)
~/.config/opencode/skills/     # OpenCode skills (also reads ~/.claude/skills/ and ~/.agents/skills/)
~/.cursor/skills/              # Cursor skills (also reads ~/.claude/skills/)
~/.kiro/skills/                # Kiro skills
~/.codebuddy/skills/           # CodeBuddy skills
~/.openclaw/skills/            # OpenClaw skills
~/.trae/skills/                # Trae skills
~/.qoder/skills/               # Qoder skills
~/.qclaw/skills/               # QClaw skills
~/.workbuddy/skills/           # WorkBuddy skills
~/.gemini/antigravity/skills/  # Antigravity skills

Key rules:

  • The real copy of every skill lives in ~/.agents/skills/<name>/
  • Each agent gets a symlink pointing to the canonical directory
  • Example: ~/.claude/skills/my-skill → ~/.agents/skills/my-skill
  • Some agents can read other agents' directories (cross-directory inheritance):
    • Codex reads ~/.agents/skills/ natively (no symlink needed)
    • Copilot reads ~/.claude/skills/
    • OpenCode reads ~/.claude/skills/ and ~/.agents/skills/
    • Cursor reads ~/.claude/skills/

Default Agent Configuration

When installing a skill, you need to decide which agents to assign it to. The recommended approach:

  1. Auto-detect installed agents — check which CLI commands exist on the system
  2. Assign to all detected agents — unless the user specifies a subset

The following script detects installed agents and builds a list of their skills directories:

bash
# Agent detection map: CLI command → skills directory
declare -A AGENT_DIRS=(
  ["claude"]="$HOME/.claude/skills"
  ["codex"]="$HOME/.codex/skills"
  ["gemini"]="$HOME/.gemini/skills"
  ["opencode"]="$HOME/.config/opencode/skills"
  ["antigravity"]="$HOME/.gemini/antigravity/skills"
  ["cursor"]="$HOME/.cursor/skills"
  ["kiro"]="$HOME/.kiro/skills"
  ["codebuddy"]="$HOME/.codebuddy/skills"
  ["openclaw"]="$HOME/.openclaw/skills"
  ["trae"]="$HOME/.trae/skills"
  ["qoder"]="$HOME/.qoder/skills"
  ["qclaw"]="$HOME/.qclaw/skills"
  ["workbuddy"]="$HOME/.workbuddy/skills"
)

# Detect which agents are installed
INSTALLED_DIRS=()
for CMD in "${!AGENT_DIRS[@]}"; do
  if which "$CMD" &>/dev/null; then
    INSTALLED_DIRS+=("${AGENT_DIRS[$CMD]}")
    echo "Detected: $CMD → ${AGENT_DIRS[$CMD]}"
  fi
done

# Copilot CLI is a GitHub CLI extension (gh copilot), not a standalone binary.
# `which gh` only confirms GitHub CLI is installed, not Copilot specifically.
# Check for the copilot extension explicitly.
if which gh &>/dev/null && gh extension list 2>/dev/null | grep -q copilot; then
  INSTALLED_DIRS+=("$HOME/.copilot/skills")
  echo "Detected: copilot → $HOME/.copilot/skills"
fi

echo "Found ${#INSTALLED_DIRS[@]} installed agents"

Use INSTALLED_DIRS as the default target when installing. If the user says "only install for Claude Code", filter to just $HOME/.claude/skills.

Lock File Format

Location: ~/.agents/.skill-lock.json (version 3)

json
{
  "version": 3,
  "skills": {
    "skill-name": {
      "source": "owner/repo",
      "sourceType": "github",
      "sourceUrl": "https://github.com/owner/repo.git",
      "skillPath": "skills/skill-name/SKILL.md",
      "skillFolderHash": "abc123...",
      "installedAt": "2024-01-01T00:00:00Z",
      "updatedAt": "2024-01-01T00:00:00Z"
    }
  },
  "dismissed": {},
  "lastSelectedAgents": []
}

Fields:

  • source: Repository identifier (e.g., "crossoverJie/skills") or local path
  • sourceType: "github", "local", or "clawhub"
  • sourceUrl: Full git URL
  • skillPath: Relative path to SKILL.md within the source
  • skillFolderHash: Git tree hash for update detection (can be empty string for local)
  • installedAt / updatedAt: ISO 8601 timestamps

Helper Functions

These reusable snippets are referenced in the operations below.

List all agent skills directories

bash
# All 14 agent skills directories (for symlink scanning)
ALL_AGENT_SKILLS=(
  "$HOME/.claude/skills"
  "$HOME/.codex/skills"
  "$HOME/.gemini/skills"
  "$HOME/.copilot/skills"
  "$HOME/.config/opencode/skills"
  "$HOME/.gemini/antigravity/skills"
  "$HOME/.cursor/skills"
  "$HOME/.kiro/skills"
  "$HOME/.codebuddy/skills"
  "$HOME/.openclaw/skills"
  "$HOME/.trae/skills"
  "$HOME/.qoder/skills"
  "$HOME/.qclaw/skills"
  "$HOME/.workbuddy/skills"
)

Find which agents have a skill installed

bash
# Usage: find_skill_agents "my-skill" "$HOME/.agents/skills/my-skill"
find_skill_agents() {
  local SKILL_NAME="$1"
  local CANONICAL="$2"
  for DIR in "${ALL_AGENT_SKILLS[@]}"; do
    local LINK="$DIR/$SKILL_NAME"
    if [ -L "$LINK" ]; then
      local RESOLVED=$(python3 -c "import os; print(os.path.realpath('$LINK'))")
      if [ "$RESOLVED" = "$CANONICAL" ]; then
        echo "$DIR"
      fi
    fi
  done
}

Create symlinks for multiple agents

bash
# Usage: create_symlinks "my-skill" "$HOME/.agents/skills/my-skill" "${TARGET_DIRS[@]}"
create_symlinks() {
  local SKILL_NAME="$1"
  local CANONICAL="$2"
  shift 2
  local TARGET_DIRS=("$@")
  for DIR in "${TARGET_DIRS[@]}"; do
    mkdir -p "$DIR"
    if [ ! -L "$DIR/$SKILL_NAME" ]; then
      ln -s "$CANONICAL" "$DIR/$SKILL_NAME"
      echo "  Symlink: $DIR/$SKILL_NAME"
    fi
  done
}

Remove symlinks from multiple agents

bash
# Usage: remove_symlinks "my-skill" "$HOME/.agents/skills/my-skill"
remove_symlinks() {
  local SKILL_NAME="$1"
  local CANONICAL="$2"
  for DIR in "${ALL_AGENT_SKILLS[@]}"; do
    local LINK="$DIR/$SKILL_NAME"
    if [ -L "$LINK" ]; then
      local RESOLVED=$(python3 -c "import os; print(os.path.realpath('$LINK'))")
      if [ "$RESOLVED" = "$CANONICAL" ]; then
        rm "$LINK"
        echo "  Removed: $LINK"
      fi
    fi
  done
}

Operations

1. Install a Skill from GitHub

Steps:

  1. Clone the repository to a temporary directory
  2. Locate the SKILL.md file and the skill directory
  3. Detect installed agents (or use user-specified targets)
  4. Copy the skill directory to ~/.agents/skills/<name>/
  5. Create symlinks for target agents
  6. Update the lock file
bash
# === Configuration ===
SKILL_NAME="my-skill"
REPO_SOURCE="owner/repo"
REPO_URL="https://github.com/owner/repo.git"
SKILL_FOLDER_PATH="skills/$SKILL_NAME"  # path to skill directory within repo

# === 1. Clone repo ===
TMPDIR=$(mktemp -d)
git clone --depth 1 "$REPO_URL" "$TMPDIR/repo"

# === 2. Locate skill directory ===
# Two path concepts (matching SkillManager's convention):
#   SKILL_FOLDER_PATH — directory containing SKILL.md (for git tree hash)
#   SKILL_FILE_PATH   — path to SKILL.md itself (for lock file's skillPath)
# Examples:
#   Sub-directory: folder="skills/foo", file="skills/foo/SKILL.md"
#   Root-level:    folder="",            file="SKILL.md"
SOURCE_DIR="$TMPDIR/repo/$SKILL_FOLDER_PATH"
if [ ! -f "$SOURCE_DIR/SKILL.md" ]; then
  # Fallback: search for SKILL.md
  FOUND=$(find "$TMPDIR/repo" -name "SKILL.md" -maxdepth 3 | head -1)
  if [ -n "$FOUND" ]; then
    SOURCE_DIR=$(dirname "$FOUND")
    # Derive folder path as relative path from repo root (empty string for root-level)
    SKILL_FOLDER_PATH="${SOURCE_DIR#$TMPDIR/repo/}"
    # Remove trailing slash if present (root case: "$TMPDIR/repo/" → "")
    SKILL_FOLDER_PATH="${SKILL_FOLDER_PATH%/}"
  else
    echo "❌ SKILL.md not found in repository"
    rm -rf "$TMPDIR"
    exit 1
  fi
fi

# Derive SKILL_FILE_PATH for lock entry: folder path + "/SKILL.md", or just "SKILL.md" at root
if [ -z "$SKILL_FOLDER_PATH" ]; then
  SKILL_FILE_PATH="SKILL.md"
else
  SKILL_FILE_PATH="$SKILL_FOLDER_PATH/SKILL.md"
fi

# === 3. Check not already installed ===
CANONICAL="$HOME/.agents/skills/$SKILL_NAME"
if [ -d "$CANONICAL" ]; then
  echo "❌ Skill '$SKILL_NAME' already exists. Use update instead."
  rm -rf "$TMPDIR"
  exit 1
fi

# === 4. Detect target agents (auto-detect all installed) ===
declare -A AGENT_DIRS=(
  ["claude"]="$HOME/.claude/skills"
  ["codex"]="$HOME/.codex/skills"
  ["gemini"]="$HOME/.gemini/skills"
  ["opencode"]="$HOME/.config/opencode/skills"
  ["antigravity"]="$HOME/.gemini/antigravity/skills"
  ["cursor"]="$HOME/.cursor/skills"
  ["kiro"]="$HOME/.kiro/skills"
  ["codebuddy"]="$HOME/.codebuddy/skills"
  ["openclaw"]="$HOME/.openclaw/skills"
  ["trae"]="$HOME/.trae/skills"
  ["qoder"]="$HOME/.qoder/skills"
  ["qclaw"]="$HOME/.qclaw/skills"
  ["workbuddy"]="$HOME/.workbuddy/skills"
)

TARGET_DIRS=()
for CMD in "${!AGENT_DIRS[@]}"; do
  if which "$CMD" &>/dev/null; then
    TARGET_DIRS+=("${AGENT_DIRS[$CMD]}")
  fi
done

# Copilot CLI is a gh extension, not a standalone binary — check explicitly
if which gh &>/dev/null && gh extension list 2>/dev/null | grep -q copilot; then
  TARGET_DIRS+=("$HOME/.copilot/skills")
fi

echo "Installing for ${#TARGET_DIRS[@]} agents: ${TARGET_DIRS[*]}"

# === 5. Copy to canonical directory ===
mkdir -p "$HOME/.agents/skills"
cp -R "$SOURCE_DIR" "$CANONICAL"

# === 6. Create symlinks ===
for DIR in "${TARGET_DIRS[@]}"; do
  mkdir -p "$DIR"
  ln -s "$CANONICAL" "$DIR/$SKILL_NAME"
  echo "  Symlink: $DIR/$SKILL_NAME"
done

# === 7. Update lock file ===
cd "$TMPDIR/repo"
# SKILL_FOLDER_PATH is empty for root-level skills; git rev-parse HEAD: returns root tree hash
TREE_HASH=$(git rev-parse HEAD:$SKILL_FOLDER_PATH 2>/dev/null || echo "")
NOW=$(date -u +"%Y-%m-%dT%H:%M:%SZ")

# Export variables so python3 can read them via os.environ
export SKILL_NAME REPO_SOURCE REPO_URL SKILL_FILE_PATH TREE_HASH NOW
python3 << 'PYEOF'
import json, os

lock_path = os.path.expanduser("~/.agents/.skill-lock.json")
os.makedirs(os.path.dirname(lock_path), exist_ok=True)

lock = {}
if os.path.exists(lock_path):
    with open(lock_path) as f:
        lock = json.load(f)

lock.setdefault("version", 3)
lock.setdefault("skills", {})
lock.setdefault("dismissed", {})
lock.setdefault("lastSelectedAgents", [])

skill_name = os.environ["SKILL_NAME"]
repo_source = os.environ["REPO_SOURCE"]
repo_url = os.environ["REPO_URL"]
# SKILL_FILE_PATH is the relative path to SKILL.md (e.g. "skills/foo/SKILL.md" or "SKILL.md")
skill_file_path = os.environ["SKILL_FILE_PATH"]
tree_hash = os.environ["TREE_HASH"]
now = os.environ["NOW"]

lock["skills"][skill_name] = {
    "source": repo_source,
    "sourceType": "github",
    "sourceUrl": repo_url,
    "skillPath": skill_file_path,
    "skillFolderHash": tree_hash,
    "installedAt": now,
    "updatedAt": now
}

with open(lock_path, "w") as f:
    json.dump(lock, f, indent=2, sort_keys=True)
PYEOF

# === 8. Cleanup ===
rm -rf "$TMPDIR"
echo "✅ Installed '$SKILL_NAME'"

2. Delete a Skill

⚠️ IMPORTANT: Always confirm with the user before deleting. Deletion removes the skill from ALL agents.

bash
SKILL_NAME="my-skill"
CANONICAL="$HOME/.agents/skills/$SKILL_NAME"

# 1. Check existence
if [ ! -d "$CANONICAL" ]; then
  echo "❌ Skill '$SKILL_NAME' not found"
  exit 1
fi

# 2. Show which agents have this skill
ALL_AGENT_SKILLS=(
  "$HOME/.claude/skills" "$HOME/.codex/skills" "$HOME/.gemini/skills"
  "$HOME/.copilot/skills" "$HOME/.config/opencode/skills" "$HOME/.gemini/antigravity/skills"
  "$HOME/.cursor/skills" "$HOME/.kiro/skills" "$HOME/.codebuddy/skills"
  "$HOME/.openclaw/skills" "$HOME/.trae/skills" "$HOME/.qoder/skills"
  "$HOME/.qclaw/skills" "$HOME/.workbuddy/skills"
)

echo "Skill '$SKILL_NAME' is assigned to:"
for DIR in "${ALL_AGENT_SKILLS[@]}"; do
  LINK="$DIR/$SKILL_NAME"
  if [ -L "$LINK" ]; then
    RESOLVED=$(python3 -c "import os; print(os.path.realpath('$LINK'))")
    if [ "$RESOLVED" = "$CANONICAL" ]; then
      echo "  ✓ $(basename $(dirname $DIR)) ($(dirname $DIR))"
    fi
  fi
done

# 3. ⚠️ MUST ask user to confirm before proceeding!
#    Prompt: "Are you sure you want to delete '$SKILL_NAME' from all agents? [y/N]"

# 4. Remove symlinks
for DIR in "${ALL_AGENT_SKILLS[@]}"; do
  LINK="$DIR/$SKILL_NAME"
  if [ -L "$LINK" ]; then
    RESOLVED=$(python3 -c "import os; print(os.path.realpath('$LINK'))")
    if [ "$RESOLVED" = "$CANONICAL" ]; then
      rm "$LINK"
    fi
  fi
done

# 5. Delete canonical directory
rm -rf "$CANONICAL"

# 6. Update lock file
export SKILL_NAME
python3 << 'PYEOF'
import json, os

lock_path = os.path.expanduser("~/.agents/.skill-lock.json")
if os.path.exists(lock_path):
    with open(lock_path) as f:
        lock = json.load(f)
    lock.get("skills", {}).pop(os.environ["SKILL_NAME"], None)
    with open(lock_path, "w") as f:
        json.dump(lock, f, indent=2, sort_keys=True)
PYEOF

echo "✅ Deleted '$SKILL_NAME'"

3. Update a Skill

Pull latest from GitHub, overwrite local copy, preserve symlinks.

bash
SKILL_NAME="my-skill"
CANONICAL="$HOME/.agents/skills/$SKILL_NAME"

# 1. Check installed
if [ ! -d "$CANONICAL" ]; then
  echo "❌ Not installed. Use install instead."
  exit 1
fi

# 2. Read lock entry — use python3 to safely extract values and write to a
#    unique temp file (avoids shell injection from untrusted lock file content,
#    and mktemp avoids race conditions between concurrent updates)
VARS_FILE=$(mktemp /tmp/skilldeck_vars.XXXXXX)
export SKILL_NAME VARS_FILE
python3 << 'PYEOF'
import json, os

lock_path = os.path.expanduser("~/.agents/.skill-lock.json")
with open(lock_path) as f:
    lock = json.load(f)

entry = lock.get("skills", {}).get(os.environ["SKILL_NAME"])
if not entry:
    print("ERROR: No lock entry found")
    exit(1)

source_url = entry.get("sourceUrl", "")
skill_path = entry.get("skillPath", "")

# deriveFolderPath: "SKILL.md" → "", "skills/foo/SKILL.md" → "skills/foo"
if skill_path == "SKILL.md":
    folder_path = ""
elif skill_path.endswith("/SKILL.md"):
    folder_path = skill_path[:-len("/SKILL.md")]
else:
    folder_path = skill_path

# Write values using declare -x syntax to avoid shell injection.
# Single-quote each value so metacharacters are not interpreted by bash.
vars_file = os.environ["VARS_FILE"]
with open(vars_file, "w") as f:
    for key, val in [("REPO_URL", source_url),
                     ("SKILL_FILE_PATH", skill_path),
                     ("SKILL_FOLDER_PATH", folder_path)]:
        # Escape single quotes within the value: ' → '\''
        safe = val.replace("'", "'\\''")
        f.write(f"declare -x {key}='{safe}'\n")
PYEOF

if [ ! -s "$VARS_FILE" ]; then
  echo "❌ Failed to read lock entry"
  rm -f "$VARS_FILE"
  exit 1
fi

# Source the safely-declared variables, then remove the temp file immediately
source "$VARS_FILE"
rm -f "$VARS_FILE"

# 3. Clone latest
TMPDIR=$(mktemp -d)
git clone --depth 1 "$REPO_URL" "$TMPDIR/repo"

# 4. Verify
SOURCE_DIR="$TMPDIR/repo/$SKILL_FOLDER_PATH"
if [ ! -f "$SOURCE_DIR/SKILL.md" ]; then
  echo "❌ SKILL.md not found at $SOURCE_DIR"
  rm -rf "$TMPDIR"
  exit 1
fi

# 4. Record existing symlinks
ALL_AGENT_SKILLS=(
  "$HOME/.claude/skills" "$HOME/.codex/skills" "$HOME/.gemini/skills"
  "$HOME/.copilot/skills" "$HOME/.config/opencode/skills" "$HOME/.gemini/antigravity/skills"
  "$HOME/.cursor/skills" "$HOME/.kiro/skills" "$HOME/.codebuddy/skills"
  "$HOME/.openclaw/skills" "$HOME/.trae/skills" "$HOME/.qoder/skills"
  "$HOME/.qclaw/skills" "$HOME/.workbuddy/skills"
)

EXISTING_LINKS=()
for DIR in "${ALL_AGENT_SKILLS[@]}"; do
  LINK="$DIR/$SKILL_NAME"
  if [ -L "$LINK" ]; then
    RESOLVED=$(python3 -c "import os; print(os.path.realpath('$LINK'))")
    if [ "$RESOLVED" = "$CANONICAL" ]; then
      EXISTING_LINKS+=("$DIR")
      rm "$LINK"
    fi
  fi
done

# 5. Replace
rm -rf "$CANONICAL"
cp -R "$SOURCE_DIR" "$CANONICAL"

# 6. Restore symlinks
for DIR in "${EXISTING_LINKS[@]}"; do
  ln -s "$CANONICAL" "$DIR/$SKILL_NAME"
done
echo "Restored ${#EXISTING_LINKS[@]} symlinks"

# 7. Update lock file
cd "$TMPDIR/repo"
# SKILL_FOLDER_PATH is empty for root-level skills; git rev-parse HEAD: returns root tree hash
TREE_HASH=$(git rev-parse HEAD:$SKILL_FOLDER_PATH 2>/dev/null || echo "")
NOW=$(date -u +"%Y-%m-%dT%H:%M:%SZ")

export SKILL_NAME SKILL_FILE_PATH TREE_HASH NOW
python3 << 'PYEOF'
import json, os

lock_path = os.path.expanduser("~/.agents/.skill-lock.json")
if not os.path.exists(lock_path):
    exit(0)

with open(lock_path) as f:
    lock = json.load(f)

skill_name = os.environ["SKILL_NAME"]
existing = lock.get("skills", {}).get(skill_name)

if existing:
    # Update existing entry — preserve all fields, only change hash and timestamp
    existing["skillFolderHash"] = os.environ["TREE_HASH"]
    existing["updatedAt"] = os.environ["NOW"]
    lock["skills"][skill_name] = existing
else:
    # No existing entry (e.g., manually installed skill) — write a complete entry
    # so LockEntry decoding doesn't fail on missing required fields
    print("⚠️ No existing lock entry found — creating a new one")
    now = os.environ["NOW"]
    lock["skills"][skill_name] = {
        "source": "",
        "sourceType": "local",
        "sourceUrl": "",
        "skillPath": os.environ["SKILL_FILE_PATH"],
        "skillFolderHash": os.environ["TREE_HASH"],
        "installedAt": now,
        "updatedAt": now
    }

with open(lock_path, "w") as f:
    json.dump(lock, f, indent=2, sort_keys=True)
PYEOF

rm -rf "$TMPDIR"
echo "✅ Updated '$SKILL_NAME'"

4. List Installed Skills

bash
echo "Installed skills:"
for DIR in "$HOME/.agents/skills"/*/; do
  [ -f "$DIR/SKILL.md" ] || continue
  NAME=$(basename "$DIR")
  DESC=$(sed -n '/^---$/,/^---$/{ /^description:/{ s/^description: *//; p; q; } }' "$DIR/SKILL.md" 2>/dev/null)
  echo "  - $NAME: $DESC"
done

5. Assign / Unassign Skill to Agent

bash
SKILL_NAME="my-skill"
AGENT_SKILLS_DIR="$HOME/.claude/skills"  # change to target agent
CANONICAL="$HOME/.agents/skills/$SKILL_NAME"

# Assign
mkdir -p "$AGENT_SKILLS_DIR"
[ -L "$AGENT_SKILLS_DIR/$SKILL_NAME" ] || ln -s "$CANONICAL" "$AGENT_SKILLS_DIR/$SKILL_NAME"

# Unassign
[ -L "$AGENT_SKILLS_DIR/$SKILL_NAME" ] && rm "$AGENT_SKILLS_DIR/$SKILL_NAME"

6. Search for Skills

bash
# skills.sh registry
curl -s "https://skills.sh/api/search?q=your-query" | python3 -m json.tool

# ClawHub
curl -s "https://api.clawhub.com/v1/skills/search?q=your-query" | python3 -m json.tool

Important Notes

  1. SkillDeck auto-refreshes: The GUI watches ~/.agents/skills/ and all agent directories. Changes via shell are reflected in the GUI within 0.5s.

  2. Lock file compatibility: Format is v3 (shared with npx skills). Always preserve version: 3 and existing entries.

  3. Cross-directory inheritance: Symlinks in ~/.claude/skills/ are also visible to Copilot, OpenCode, and Cursor. Don't create redundant symlinks.

  4. Codex special case: Reads ~/.agents/skills/ natively. Symlink in ~/.codex/skills/ is optional but consistent.

  5. Never modify: ~/.agents/.skilldeck-cache.json (SkillDeck private cache)

Frequently asked questions

What does the Skilldeck AI skill do?

Manage AI agent skills via natural language — install, update, delete, and assign skills to agents using SkillDeck's filesystem conventions.

Why use Skilldeck on TypingMind?

Because you install it once and use it with any model. Skilldeck is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Skilldeck in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/crossoverJie/SkillDeck/tree/main/skills/skilldeck. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Skilldeck?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Skilldeck?

As many as you like. As long as a model supports skills, you can use Skilldeck with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Skilldeck AI skill free?

Yes. It is published on GitHub by crossoverJie under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

Related AI Skills

View all
openclaw logo

Clawdtributor

openclaw
OrganizationPopular

Clawtributor PRs here, last week or another window: discover conversation refs, recheck GitHub, rank by impact.

82.2K
391K
Instructions
View
openclaw logo

Clawsweeper

openclaw
OrganizationPopular

Use for all ClawSweeper work: OpenClaw issue/PR sweep reports, repair jobs, cloud fix PRs, @clawsweeper maintainer mention commands, trusted ClawSweeper-reviewed autofix/automerge, GitHub Actions…

82.2K
391K
1 files
View
openclaw logo

Control Ui E2e

openclaw
OrganizationPopular

Use when designing, testing, fixing, or extending the OpenClaw Control UI GUI, including UI stress-test galleries with feedback inputs, Vitest + Playwright end-to-end checks, mocked Gateway flows,…

82.2K
391K
1 files
View
openclaw logo

Deslop

openclaw
OrganizationPopular

Diff-scoped AI-slop cleanup pass: strip comment slop, defensive-check slop, type-laundering, and style drift from the current branch diff before autoreview.

82.2K
391K
Instructions
View

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇