drop-close

name	drop-close
description	Close Pulse worker (Drop) threads. Writes structured deposit JSON for orchestrator review. Does NOT commit - that's the orchestrator's job. For normal threads use thread-close. For post-build synthesis use build-close.

Drop Close

Close Pulse Drop worker threads with structured deposits.

When to Use

• Pulse Drops (headless workers spawned via /zo/ask)
• Any thread with drop_id in SESSION_STATE

Wrong tool?

• If normal thread → use thread-close
• If closing a completed build → use build-close

Quick Start

# Close current drop (reads context from SESSION_STATE)
python3 Skills/drop-close/scripts/close.py --convo-id con_XXXXX

What It Does

1. Validates context — Guards check you ARE a Drop
2. Reads drop_id and build_slug from SESSION_STATE
3. Generates title — 3-slot emoji system with build context
4. Gathers deposit data:
   • Summary of work done
   • List of artifacts created
   • Key learnings
   • Concerns for orchestrator
   • Decisions made (with rationale)
   • Concepts exercised (from Learning Landscape)
5. Writes deposit JSON to N5/builds/<slug>/deposits/<drop_id>.json
6. Does NOT commit — Orchestrator handles that

Title Generation (REQUIRED)

After the script outputs context JSON, you MUST generate a title using the 3-slot emoji system.

Format

MMM DD | {state} {type} {content} [Build Title] Drop Task Summary

For Drops:

• Slot 1 (State): ✅ complete, ❌ failed, ⏸️ blocked
• Slot 2 (Type): Always 👷🏽‍♂️ (worker)
• Slot 3 (Content): Match the work type (🏗️ build, 🛠️ repair, 🔎 research, etc.)
• [Brackets]: ALWAYS include the build title (from build_slug → meta.json)

Example Titles

Jan 24 | ✅ 👷🏽‍♂️ 🏗️ [Pulse System] D1.1 Core Script Implementation
Jan 24 | ✅ 👷🏽‍♂️ 🛠️ [Close Skills] D2.3 Guards Module Fixes
Jan 24 | ❌ 👷🏽‍♂️ 🏗️ [API Integration] D1.2 Auth Handler (blocked on API key)

Title Generation Steps

1. Get build_slug from SESSION_STATE or script output
2. Look up build title: cat N5/builds/<slug>/meta.json | jq -r '.title'
3. Determine state from work outcome (complete/failed/blocked)
4. Use 👷🏽‍♂️ for type (always worker)
5. Pick content emoji based on what the drop did
6. Include drop_id (D1.1, D2.3, etc.) in the semantic title
7. Describe what was accomplished in 3-5 words

Emoji Quick Reference

Slot	Emoji	Meaning
State	✅	Complete
State	❌	Failed
State	⏸️	Blocked/waiting
Type	👷🏽‍♂️	Worker (always for drops)
Content	🏗️	Building/implementing
Content	🛠️	Fixing/repairing
Content	🔎	Research/investigation
Content	✍️	Writing/content
Content	📊	Data work

Deposit Format

{
  "drop_id": "D1.1",
  "convo_id": "con_XXXXX",
  "status": "complete",
  "completed_at": "2026-01-24T12:00:00Z",
  "summary": "Created foo.py and bar.py with tests",
  "artifacts": [
    "src/foo.py",
    "src/bar.py"
  ],
  "learnings": "The API requires auth header even for public endpoints",
  "concerns": "Rate limiting may be an issue at scale",
  "decisions": [
    {
      "decision": "Used async/await over threads",
      "rationale": "Better performance for I/O-bound operations"
    }
  ],
  "concepts_exercised": ["async-patterns", "error-handling"]
}

Options

• --convo-id (required): Conversation ID
• --force: Bypass context guards
• --status: Override status (complete|partial|failed|blocked)

Fail-Safes

This skill includes context guards. If you call it incorrectly:

⚠️  WRONG SKILL DETECTED

You called: drop-close
Suggested:  thread-close
Reason:     No drop context — use thread-close instead

Run the suggested skill instead, or use --force to override.

Integration with Pulse

Drops should call this at the end of their work:

# In drop brief template
# At end of work:
# python3 Skills/drop-close/scripts/close.py --convo-id {CONVO_ID}

The deposit is then:

1. Detected by Pulse tick loop
2. Filtered via LLM judgment
3. Used in build-close synthesis

Checklist Before Completing

• Title generated with all 3 emoji slots
• Title includes [Build Title] in brackets
• Title includes drop_id (D#.#)
• Deposit JSON written to correct path
• Artifacts list is complete and uses relative paths
• Learnings captured (if any discoveries made)
• Concerns noted (if any issues for orchestrator)
• Concepts exercised listed (if build has Learning Landscape)

Related

• file 'N5/config/emoji-legend.json' — Full emoji definitions
• file 'N5/lib/close/emoji.py' — Title generation helpers
• file 'Skills/thread-close/SKILL.md' — Normal thread close
• file 'Skills/build-close/SKILL.md' — Build synthesis close