菜单
Set up a polling loop on a card where you are waiting for another agent to respond. Auto-blocks the card after a timeout with no response.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Session start checklist. Runs inbox check, board state review, knowledge freshness audit, and intent declaration before pulling any work. The session counterpart to /lets-wrap.
End-of-session wrap-up checklist. Prompts the agent to follow through on knowledge capture, board hygiene, and memory updates before signing off. Use when ending a working session.
Security penetration test + GDPR/SOC 2/ISO 27001 compliance audit. Targets a single project or the estate. Produces categorised findings, board cards, and regression tests. Invocable by any agent or the product owner.
Agent-only skill. After feature tests pass, walks the agent through regression test vectors for any new or modified API routes. Produces regression tests and documents coverage on the card. Invoked by the regression-gate hook, not by humans.
Exploratory testing from a user's perspective. AI-reasoned edge cases, happy path validation, non-happy-path scenarios, regression quality assessment. The closest thing to having a human test team. Invocable by any agent or the product owner.
| name | watch-card |
| description | Set up a polling loop on a card where you are waiting for another agent to respond. Auto-blocks the card after a timeout with no response. |
| disable-model-invocation | true |
Use this immediately after posting a routing comment on a card and needing to wait for another agent to respond.
/watch-card <card_id> <prefix_you_notified> <your_prefix>
Example: you commented [ProjectAgent] on card 123, your prefix is Agent-Orchestrator:
/watch-card 123 ProjectAgent Agent-Orchestrator
Polls your inbox every 10 minutes for a response referencing #<card_id>. After 6 attempts (60 minutes) with no reply, blocks the card and adds a comment for the product owner.
Step 1 — Create an attempt counter file immediately:
echo 0 > /tmp/watch-<card_id>.count
Step 2 — Create a cron job to run every 10 minutes. The command:
COUNT=$(cat /tmp/watch-<card_id>.count 2>/dev/null || echo 0)
COUNT=$((COUNT + 1))
echo $COUNT > /tmp/watch-<card_id>.count
if board-tool inbox <your_prefix> | grep -q "#<card_id>"; then
echo "Response received for #<card_id> - cancelling watch"
rm -f /tmp/watch-<card_id>.count
# Cancel this cron by its ID (note the cron ID when you create it)
else
if [ "$COUNT" -ge 6 ]; then
board-tool block <card_id> "Waiting for [<prefix_notified>] response - no reply after 1hr"
board-tool comment <card_id> "[watch-card] No response from [<prefix_notified>] after 1hr. Card blocked. PO: please check inbox or prompt the relevant agent."
rm -f /tmp/watch-<card_id>.count
# Cancel this cron by its ID
else
echo "Attempt $COUNT/6 - no response yet for #<card_id>"
fi
fi
Step 3 — Note the cron ID returned by the cron tool. Use it to cancel the loop once the response arrives (or once it auto-blocks).
Step 4 — Continue your own work. The loop fires every 10 minutes and terminates itself at 6 attempts.
| Agent type | Default poll | Active wait |
|---|---|---|
| Project agents | 1hr + jitter | /watch-card (10min, 6 attempts) |
| Orchestrator / specialists | 15min | /watch-card (10min, 6 attempts) |
You are in active dialogue with another agent. The other agent's inbox poll interval determines response latency; orchestrator and specialist agents poll every 15 minutes, project agents every hour. Polling your own inbox at 10 minutes means you act on a response within one check of receiving it, keeping dialogue latency manageable.