skiphow-longrun git:20260925.706ca87A

Immutable. This exact content never changes and is served at /api/v1/blob/aa8d24e9000f34e8.

---
name: skiphow-longrun
description: Deliver an accepted epic, milestone, or issue set through delegated implementation and independent review loops. Use for sustained execution and resumes of tracked work, with CTO sequencing, integration, and process diagnosis.
---

# SkipHow longrun

Deliver every authorized outcome in the requested work set to its agreed destination. Before consequential work, have the [SkipHow CTO kernel](../skiphow/SKILL.md) in context. Read it if absent. The lead remains the accountable CTO throughout execution.

Use [tracked work](../skiphow/references/tracked-work.md) to reconcile current records, accepted decisions, live branches, ownership, and prior delivery before starting or resuming. Retire superseded instructions in execution records no live session is writing. Leave a live session's lane to it and continue independent authorized work around it and around owner decisions.

Use [delegation](../skiphow/references/delegation.md) to size and sequence verifiable slices. Delegate implementation and independent review to fresh agents for each slice, retaining its implementer and reviewer through corrections where the host supports continuation. If a replacement is necessary, carry forward the slice's findings and evidence. Keep scope, product decisions, synthesis, integration, and the completion claim with the CTO.

Apply the shared model and effort routing rule at every dispatch through actual host controls. Reassess routing or split the assignment when a miss, growing scope, or context pressure makes its current shape unreliable. Parallelize ready independent work within verified isolation and integration capacity. Without verified writer isolation, explain the limitation and keep the lead as the only writer while delegates investigate and independently review. If independent review itself is unavailable, report that slice's review blocker rather than claiming the loop complete.

Use [verification](../skiphow/references/verification.md) for the fix/review loop. Check returned evidence yourself, confirm findings, have the implementer repair qualifying defects, and obtain targeted independent review of the corrections. Revalidate affected evidence before accepting a slice into the lead's working state. Use [integration](../skiphow/references/integration.md) to combine accepted results, verify them together, and choose delivery units. A slice's handoff to the lead and the owner's delivery can have different destinations. Close each record only when its own promised outcome holds under [tracked work](../skiphow/references/tracked-work.md#closing-what-the-tracker-carries); accepting a handoff does not close an outcome still awaiting delivery. An implementation report or a review verdict alone does not establish delivery.

Use [diagnosis](../skiphow/references/diagnosis.md) when a test, delegate, tool, or integration behaves anomalously, progress loses its signal, or the same failure returns. Use [operations](../skiphow/references/operations.md) when the cause belongs to the delivery system. Diagnose the responsible layer, change direction when evidence warrants it, and keep independent outcome-producing work moving.

Checkpoint through the existing tracking and continuation conventions at meaningful boundaries. Continue beyond slice handoffs until every authorized outcome is verified at its agreed destination or has a concrete recorded blocker. Report delivered outcomes and evidence, accepted results still awaiting delivery, remaining review or owner actions, and owned resources that could not be retired. Production and public releases retain the kernel's explicit-grant boundary.