القائمة
Triage incoming tasks, alerts, and events into P0/P1/P2 buckets. Use whenever Strategist receives a new task, an unhandled alert, or is deciding whether to issue a directive.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | priority-triage |
| description | Triage incoming tasks, alerts, and events into P0/P1/P2 buckets. Use whenever Strategist receives a new task, an unhandled alert, or is deciding whether to issue a directive. |
| metadata | {"version":"1.0.0","type":"always-active"} |
Classify every incoming task, alert, or event into one of three buckets before issuing a directive or escalating. Clear bucket assignment prevents priority inflation.
| Trigger | Example |
|---|---|
| Production outage | Agents API returning 500s, Mission Control unreachable, event bus down |
| Security incident | Credential leak, unauthorized action detected, safety gate bypass |
| Human-escalated urgent | Elon pings #yclaw-alerts with "🚨 now" or direct mention with urgency |
| Legal/compliance trigger | Outbound content flagged for securities language, DMCA notice |
| Data loss imminent | Agent writing unreviewed content to a public channel, deploy rolling back inflight |
Rule: Any action required to prevent or stop immediate damage. If you're not sure whether it's truly P0, treat it as P1 until evidence of real-world impact arrives.
| Trigger | Example |
|---|---|
| Blocked agent task | Architect reports a task blocked >4 hours, Builder DLQ retry exhausted |
| Stale review queue | Reviewer's reviewer:queue_stale event fires (>5 pending or >4h old) |
| Repeated agent failure | Same task failed 2+ times today (next fail = escalate to P0) |
| Cross-agent coordination gap | Ember needs an asset Forge hasn't produced; deadline within 24h |
| Stale PR | PR on YClawAI/YClaw CI-green + unreviewed for >1 hour |
Rule: Max 5 P1 items active at once. If you're adding a 6th, something must be demoted to P2 or resolved first. Too many P1s means nothing is actually P1.
| Trigger | Example |
|---|---|
| Optimization idea | "We could cache X", "Ember could batch posts better" |
| Audit finding (non-critical) | Prompt contains a minor inconsistency, doc says "Slack" but should say "Discord" |
| Nice-to-have feature | Adding a new agent that isn't required by org mission |
| Cost trend concern | LLM spend up 15% month-over-month but still under budget |
| Recurring minor issue | Same agent posts slightly long messages occasionally |
Rule: Never dispatch a directive for a P2 item mid-heartbeat. Collect them for the next scheduled planning cycle. Acting on P2 items out-of-cycle creates noise.
Ask these in order. Stop at the first YES.
If unsure between P0 and P1 → default P1. (You can always re-classify upward once evidence arrives.) If unsure between P1 and P2 → default P2. (Inflating P1 dilutes urgency.)
Scenario: Builder DLQ has 3 entries, all for the same repo, all failing CI.
strategist:architect_directive with the 3 correlation IDs and ask
for a single plan to unblock the pattern.Scenario: sentinel:alert fires — Atlas connection dropping intermittently.
Scenario: Ember's content score averaged 72/100 this week (down from 85).
escalation-triage skill.directive-authoring skill.standup-synthesis skill.Detect and prevent product-specific terminology contamination across skills/ and prompts/. Run after any bulk content update, skill import from another project, or agent config change. Applied as a shared skill across all agents.
When and how Architect delegates tasks to AO, Mechanic, and Designer.
Checklist and rules for reviewing deployment requests. Architect assesses but NEVER executes.
Rules for triaging new issues, classifying severity, and routing to the correct agent.
How to classify and respond to CI/pipeline failures.
Rules for detecting and managing stale issues and PR reviews.