close

name	close
description	Universal close skill. Just say "close" and it auto-routes to the right close skill (thread-close, drop-close, or build-close) based on SESSION_STATE context. Works in both native Zo and Claude Code environments.
compatibility	Created for Zo Computer
metadata	{"author":"<YOUR_HANDLE>.zo.computer"}

Close

Step 1: Detect Environment

Check if running in Claude Code:

python3 N5/scripts/cc_close_bridge.py detect

• If CLAUDE_CODE=false → Native Zo path (Step 2a)
• If CLAUDE_CODE=true → Claude Code path (Step 2b)

Step 2a: Native Zo

Run the router with the current conversation ID:

python3 Skills/thread-close/scripts/router.py --convo-id <CONVO_ID>

The router reads SESSION_STATE and picks the right skill.

Step 2b: Claude Code

Claude Code sessions don't have a Zo conversation ID or SESSION_STATE. Use the bridge to create a synthetic workspace first.

You (the LLM) must determine these from conversation context:

1. type — What kind of work was this? (build, research, discussion, planning, debug)
2. focus — One-line summary of what the session was about
3. tier — Assess complexity:
   • Tier 1: Simple discussion, < 3 artifacts
   • Tier 2: Standard work, 3-10 artifacts, research
   • Tier 3: Builds, complex multi-file work, orchestration
4. build-slug (optional) — If this was work on a Pulse build
5. artifacts (optional) — Key files created or modified

Then run:

python3 N5/scripts/cc_close_bridge.py init
  --type <type>
  --focus "<focus>"
  --tier <tier>

Add --build-slug <slug> if build context applies. Add --artifacts path1 path2 ... for key files.

Capture the CC_CONVO_ID from output, then run the normal router:

python3 Skills/thread-close/scripts/router.py --convo-id <CC_CONVO_ID>

Step 3: Follow Thread-Close Instructions

From here the flow is identical in both environments. Read and follow file Skills/thread-close/SKILL.md (or whichever skill the router selected).

The script outputs JSON context → you do semantic analysis → you call write functions → you echo the title.