| name | ELTONSLAPTOP_QWEN-dropchat-multipc |
| description | Push session summary to cross-PC gateway with peer ID "ELTONSLAPTOP_QWEN". Triggers — /dropchat-multipc, "drop chat to multi-pc", "broadcast session summary". Aliases — dropchat, dctm, multipc-sync, peer-sync. |
/dropchat-multipc — Qwen edition (Peer ID: ELTONSLAPTOP_QWEN)
End-of-session handoff for cross-PC agent coordination. Compresses session deliverables into structured broadcast, sends via gateway, drains peer inbox.
Peer ID: ELTONSLAPTOP_QWEN
When to use
- End of long autonomous session (10+ tool calls)
- Before pausing work for the day
- User says "/dropchat-multipc", "drop chat to multi-pc", "share with peers"
- After landing changes touching code another agent might modify
Prerequisites
Gateway health check
Always use 192.168.2.32:8788 — never 127.0.0.1.
curl.exe -s -m 5 http://192.168.2.32:8788/health | python -m json.tool
Core files exist
if (Test-Path 'tools\adapters\cursor_claude_adapter.py') { echo "adapter OK" } else { echo "adapter MISSING" }
if (Test-Path 'tools\protocol_inspect.py') { echo "inspect OK" } else { echo "inspect MISSING" }
Step 1 — Gather payload
Build structured JSON from real deliverables:
{
"schema": "session-summary/v1",
"agent": "qwen-code-desktop",
"peer_id": "ELTONSLAPTOP_QWEN",
"branch_at_end": "<current git branch>",
"commits_this_session": [{"sha": "<short>", "subject": "..."}],
"files_written_or_modified": ["<path>", ...],
"decisions": ["1-line decision summary"],
"open_questions": ["1-line question"],
"follow_ups_for_other_agents": [
{"to": "all", "ask": "...", "priority": "P0|P1|P2"}
]
}
Step 2 — Send broadcast
python tools/adapters/cursor_claude_adapter.py \
--peer-id ELTONSLAPTOP_QWEN \
--http-base http://192.168.2.32:8788 \
send --topic SESSION_SUMMARY \
--payload <your-JSON-payload> \
--to all
Step 3 — Verify send
python tools/protocol_inspect.py tail --limit 5
Should see your envelope near top.
Step 4 — Drain inbox
Two pulls:
python tools/adapters/cursor_claude_adapter.py \
--peer-id ELTONSLAPTOP_QWEN \
--http-base http://192.168.2.32:8788 \
poll --limit 50 > /tmp/inbox_dms.json
python tools/adapters/cursor_claude_adapter.py \
--peer-id all \
--http-base http://192.168.2.32:8788 \
poll --limit 50 > /tmp/inbox_broadcasts.json
Polling is destructive — messages pop off queue. Durable copy at logs/cross_pc_protocol/events.jsonl.
Step 5 — Triage peer messages
For each message:
- Classify by topic (
SESSION_SUMMARY, task.request, audit_finding, etc.)
- If P0 bug overlapping your session's work → re-verify on disk before acting
- If
require_ack=true → POST /ack after handling
- Log to
reports/peer_inbox_<UTC>.md
Step 6 — Closing broadcast (optional)
python tools/adapters/cursor_claude_adapter.py \
--peer-id ELTONSLAPTOP_QWEN \
--http-base http://192.168.2.32:8788 \
send --topic SESSION_CLOSED \
--payload '{"peer":"ELTONSLAPTOP_QWEN","drained_n":<count>,"acted_on":<count>}' \
--to all
Output format
Compact summary:
[dropchat-multipc] <timestamp UTC>
broadcast: SESSION_SUMMARY → all (message_id=<id>)
inbox: <n> DMs, <m> broadcasts pulled
peers reachable: <count from /health>
revised: 0
Failure logging
If bus unreachable at any point:
- Append entry to
CHATBIBLE_FAILURE.MD
- Commit + push
CHATBIBLE_FAILURE.MD to main
Silent failures cause multi-agent divergence. Log every true unreachable-bus event.
Anti-patterns
- Do NOT post raw chat transcripts — summarize to structured payloads
- Do NOT broadcast on every tool call — once per session or when invoked
- Do NOT delete
logs/cross_pc_protocol/events.jsonl — replay log
- Do NOT use
127.0.0.1 — always use 192.168.2.32 on this desktop