Skill Push

Skill Push

Create and push reusable skills to the SkillNote registry so all connected agents learn from repeated patterns.

The SkillNote API is at: http://${CLAUDE_PLUGIN_OPTION_HOST:-localhost}:8082

When to Act

• The user gave the same instruction or convention 2+ times this session
• The user explicitly asks to create, save, or push a skill
• Session retrospective — review for saveable patterns before wrapping up

Only suggest for PERSISTENT conventions, not temporary workarounds.

Step 1: Confirm (REQUIRED — never skip)

If the user did not explicitly ask to create a skill, you MUST ask first before doing anything else.

Tell the user what you noticed in one sentence. Be specific. Then ask: "Want me to save this as a SkillNote skill?"

Stop and wait for an explicit yes. Do NOT draft the name/description/content, fetch collections, or present a preview until the user confirms. Unsolicited drafts feel pushy and waste attention.

Skip this step only when the user already said "create a skill for X" / "save this pattern" / "push a skill" in their own words.

Step 2: Draft

Collaborate on three fields:

name: lowercase, hyphens, digits only, max 64 chars. Also used as slug.

description (max 1024 chars — THIS IS THE TRIGGER): Must include what the skill does + explicit trigger keywords. Agents decide to use a skill based solely on its description.

content: Instructions under 200 lines. Start with # Title. Be actionable with examples.

Show the full draft to the user.

Step 3: Check if Exists

import urllib.request, json, os
api = f"http://{os.environ.get('CLAUDE_PLUGIN_OPTION_HOST', 'localhost')}:8082"
try:
    resp = urllib.request.urlopen(f"{api}/v1/skills/<SLUG>")
    print(f"EXISTS: v{json.loads(resp.read()).get('current_version', '?')}")
except urllib.error.HTTPError as e:
    print("NEW" if e.code == 404 else f"ERROR: {e.code}")

Step 4: Choose Collection

import urllib.request, json, os
api = f"http://{os.environ.get('CLAUDE_PLUGIN_OPTION_HOST', 'localhost')}:8082"
try:
    cols = json.loads(urllib.request.urlopen(f"{api}/v1/collections").read())
    for c in cols: print(f"  - {c['name']} ({c['count']} skills)")
    if not cols: print("  (no collections yet)")
except Exception as e: print(f"Could not fetch: {e}")

Every skill must belong to at least one collection. Use AskUserQuestion to let the user pick:

• Show existing collections from the list above as options
• Include an option to type a new collection name — names must match ^[a-z0-9_-]+$ (lowercase letters, numbers, hyphens, underscores), 1–128 chars, and cannot contain reserved words anthropic or claude
• Recommend the collection that best fits the skill's domain
• A skill cannot be pushed without a collection

If the user types an invalid name, the POST /v1/skills call will return 422 COLLECTION_NAME_INVALID. Surface the error, explain the rule, and ask them to type a valid name before retrying.

Step 5: Final Review

Show complete skill preview. Emphasize: "Does the description have good trigger keywords?" Wait for approval.

Step 6: Push

New skill:

import json, urllib.request, os
api = f"http://{os.environ.get('CLAUDE_PLUGIN_OPTION_HOST', 'localhost')}:8082"
payload = json.dumps({"name": "<NAME>", "slug": "<NAME>", "description": "<DESC>", "content_md": "<CONTENT>", "collections": ["<COL>"]}).encode()
req = urllib.request.Request(f"{api}/v1/skills", data=payload, headers={"Content-Type": "application/json"}, method="POST")
try:
    result = json.loads(urllib.request.urlopen(req).read())
    print(f"Created: {result['slug']} v{result['current_version']}")
except urllib.error.HTTPError as e: print(f"Error: {json.loads(e.read())}")

Existing skill (update):

import json, urllib.request, os
api = f"http://{os.environ.get('CLAUDE_PLUGIN_OPTION_HOST', 'localhost')}:8082"
payload = json.dumps({"name": "<NAME>", "description": "<DESC>", "content_md": "<CONTENT>", "collections": ["<COL>"]}).encode()
req = urllib.request.Request(f"{api}/v1/skills/<SLUG>", data=payload, headers={"Content-Type": "application/json"}, method="PATCH")
try:
    result = json.loads(urllib.request.urlopen(req).read())
    print(f"Updated: {result['slug']} v{result['current_version']}")
except urllib.error.HTTPError as e: print(f"Error: {json.loads(e.read())}")

After success: link to http://${CLAUDE_PLUGIN_OPTION_HOST:-localhost}:3000/skills/<slug> for viewing.

Error Reference

• 422: Name format wrong or description has XML tags
• 409: Slug exists — switch to PATCH
• Connection refused: API unreachable