| name | interrupt |
| description | Handle urgent interrupt requests. Sizes the impact, shows tradeoffs against current plan, and helps decide whether to take it on, delegate, or push back. |
| metadata | {"workbench.argument-hint":"<interrupt description>"} |
Path resolution: This skill may run from any repo. All context/ and config.yaml paths are relative to the workbench root, not the current working directory. Read ~/.codex/workbench-root or ~/.claude/workbench-root to get the absolute workbench path, then prepend it to all context/ and config.yaml references. See PATHS.md.
Interrupt Handler
Someone just said "this is urgent, it needs to happen now." This skill makes the tradeoffs visible so a good decision gets made fast.
Input
The user will describe:
- What the new urgent task is
- Who requested it (manager, team lead, stakeholder)
- Any deadline mentioned
Process
Step 1: Size the Interrupt
Break the new task into concrete subtasks and estimate each one. Apply the estimation rules:
- Get the gut feel for each subtask
- Apply the 1.5x multiplier (2x if it involves unfamiliar code, coordination with others, or unknowns)
- Add PR review cycle time (0.5 day minimum)
- Add context-switching overhead (0.5h for getting into it, 0.5h for getting back to what you were doing)
Present:
New task: [description]
Requested by: [who]
Estimated effort: X-Y hours (best case / realistic)
Subtasks:
- Subtask 1: Xh
- Subtask 2: Yh
- PR review + iteration: Zh
- Context switch overhead: 1h
Step 2: Show Current State
Read today's daily plan (context/active/daily.md) and the weekly plan (context/active/weekly.md).
Present:
Today's remaining work:
- [ ] Item A (est: Xh) — WHY: reason
- [ ] Item B (est: Yh) — WHY: reason
This week's capacity:
- Committed: Xh of Yh available
- Remaining capacity: Zh
- Already at risk: [any items close to deadline]
Step 3: Show the Impact
Make the tradeoffs explicit:
Option 1: Take it on now
-> Item A slips to [when]
-> Item B slips to [when]
-> [Person waiting on A] will be affected
-> Weekly capacity goes from X% to Y% utilized
Option 2: Take it on, but defer lower-priority work
-> Complete by [when]
-> Defer [specific items] to next week
-> Impact: [who is affected by the deferral]
Option 3: Delegate it
-> This task [does/doesn't] require principal-level expertise
-> Could be handled by: [suggestion based on complexity]
-> You'd need to: [brief handoff effort estimate]
-> Net time cost to you: Xh (handoff) vs Yh (doing it yourself)
Option 4: Push back with a counter-timeline
-> "I can start this [when] and have it done by [when]"
-> Current work stays on track
-> Risk: [whoever asked] may not accept the delay
Step 4: Help Respond
Based on the decision:
If taking it on:
- Update today's daily plan — move the new task to #1, shift everything else
- Update the weekly plan with new capacity numbers
- Flag anything that now needs to be communicated to stakeholders
If delegating:
- Draft a handoff message with context the other engineer needs
- Offer to send it via Slack
If pushing back:
- Draft a response to the requester with:
- Acknowledgment that it's important
- Your current commitments and why they matter
- A realistic counter-timeline
- What would need to change for you to do it sooner
Important Notes
- Never frame this as "you're too busy." Frame it as tradeoffs: "here's what taking this on costs."
- The goal is to make the requester's decision informed, not to say no.
- If the requester is a VP or above, note that but don't automatically capitulate — still show the tradeoffs.
- If this is a genuine emergency (production down, data loss, security issue), skip the analysis and just flag what gets deferred.
- Speed matters here. The whole analysis should take under 60 seconds to produce.