Aramb Skills

aramb-skills CLI

The aramb-skills binary is pre-installed at /usr/local/bin/aramb-skills in every agent runtime container. No installation step is needed.

Environment Variables

Both can also be passed as per-command flags (--token, --base-url), which take precedence over env vars.

Commands

search — find skills by text, category, or tag

At least one of query text, --category, or --tag is required.

# Text search
aramb-skills search "planning"

# Filter by category
aramb-skills search --category planner --limit 5

# Filter by tag with text
aramb-skills search --tag automation "summarize"

# Combined filters, JSON output
aramb-skills search --category planner --tag planning --json

Output columns: SKILL (owner/repo/name), CATEGORY, AUTHOR, STARS, DL

Flags: --category (slug), --tag (slug), --limit (default 20), --page (default 1), --json

list — list all skills with optional filters

aramb-skills list
aramb-skills list --status approved --limit 10
aramb-skills list --category automation --json

Flags: --category, --status (draft/approved/pending), --limit, --page, --json

featured — list featured skills

aramb-skills featured
aramb-skills featured --limit 5 --json

get — inspect a single skill by full ID

Full ID format: owner/repo/name

aramb-skills get clode-labs/skills-registry/planner
aramb-skills get clode-labs/skills-registry/planner --json

Outputs: name, description, category, author, license, compatibility, status, tags, popularity, updated date.

JSON output example:

{
  "success": true,
  "data": {
    "full_id": "clode-labs/skills-registry/planner",
    "name": "Planner",
    "category": "planner",
    "status": "approved",
    "tags": ["planning", "agent"],
    "author_name": "clode-labs"
  }
}

download — download a skill's files as ZIP and extract to a directory

# Extract into current directory
aramb-skills download clode-labs/skills-registry/planner

# Extract into a specific directory (created if it doesn't exist)
aramb-skills download clode-labs/skills-registry/planner --output /tmp/skill

# Machine-readable result (stdout) with progress on stderr
aramb-skills download clode-labs/skills-registry/planner --output /tmp/skill --json

# Private skill
ARAMB_API_TOKEN=eyJ... aramb-skills download my-org/private-repo/my-skill --output ./skill

JSON stdout on success: {"success":true,"output":"/tmp/skill"}

Progress goes to stderr during extraction:

Downloading clode-labs/skills-registry/planner...
  SKILL.md
  main.ts
Extracted to /tmp/skill

version — print binary version

aramb-skills version

Auth Behaviour

Exit Codes

Typical Workflow: Download a Skill into an Agent Workspace

Use this when a skill is needed in an agent workspace and not already present in {{SKILLS_SOURCE}}/.

# 1. Search the registry to find the right skill
aramb-skills search --category planner

# 2. Inspect the skill before downloading
aramb-skills get clode-labs/skills-registry/planner

# 3. Download into the agent's skills directory
SKILL_DIR="${DESTINATION}/skills/planner"
aramb-skills download clode-labs/skills-registry/planner --output "$SKILL_DIR" --json

# 4. Verify the download succeeded
[ -f "$SKILL_DIR/SKILL.md" ] || { echo "ERROR: Skill download failed — SKILL.md missing"; exit 1; }
echo "Skill installed at: $SKILL_DIR"

Checking Before Copying

Before copying a skill from {{SKILLS_SOURCE}}/, verify it exists there first. If not, fall back to downloading from the registry:

SKILL_NAME="planner"
SOURCE_DIR="{{SKILLS_SOURCE}}/${SKILL_NAME}"
DEST_DIR="${DESTINATION}/skills/${SKILL_NAME}"

if [ -d "$SOURCE_DIR" ]; then
  cp -r "$SOURCE_DIR" "$DEST_DIR"
  echo "Copied $SKILL_NAME from local source"
else
  echo "$SKILL_NAME not in local source — downloading from registry..."
  aramb-skills search "$SKILL_NAME" --limit 3
  # Inspect and confirm the right skill, then download:
  aramb-skills download clode-labs/skills-registry/"$SKILL_NAME" --output "$DEST_DIR" --json
  [ -f "$DEST_DIR/SKILL.md" ] || { echo "ERROR: Download failed"; exit 1; }
fi