# Kbd Cancel

> Use to gracefully cancel the active KBD run while preserving its checkpoint and immutable audit history. Requires a non-empty cancellation reason. Cancellation is terminal.

- Skill: `prometheus-ags/kbd-cancel` (Agent Skill)
- Install (CLI): `npx get-skill add prometheus-ags/kbd-cancel`
- Raw SKILL.md: https://api.skillmd.com/api/skills/prometheus-ags/kbd-cancel/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- License: MIT
- Author: Prometheus-AGS (https://skillmd.com/u/prometheus-ags)
- Updated: 2026-11-18
- Page: https://skillmd.com/skills/prometheus-ags/kbd-cancel

---


# /kbd-cancel

Cancel the active run as an explicit terminal operator action.

## Progress Signals (MANDATORY)

Before cancellation, emit:

```text
Starting kbd-cancel — <phase-name>
```

After durable cancellation, emit:

```text
Completed kbd-cancel — <phase-name> cancelled
```

## Procedure

1. Require a non-empty cancellation reason.
2. If this project's own runtime CLI is available, run its cancel command with
   the reason.
3. Otherwise, create `.kbd-orchestrator/PAUSE` first and atomically set the
   waypoint status to `cancelled` (via `lib/platform/atomic-write.mjs`'s
   `atomicWrite`), recording the prior status, reason, actor, and timestamp.
   Before writing, call `assertProjectionWritable(root, target)` from
   `lib/kbd/runtime-authority.mjs` — it throws when the waypoint is
   runtime-authoritative (`generatedBy === 'kbd-runtime'`), which is exactly
   the case where a direct write here would be wrong.
4. Preserve all checkpoints and dirty-work metadata.
5. Invoke a host-native cancel operation when the harness exposes one.

Cancellation is terminal. Resuming it requires a new run, never an in-place
status edit.

