> ## Documentation Index
> Fetch the complete documentation index at: https://docs.crewship.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Routine Waitpoint Commands

> Inspect and decide human approval waitpoints created by routines.

# crewship routine waitpoints

Waitpoints are HITL pauses created by a routine `wait` step of
`kind: approval`. A parked run resumes when a human approves or rejects the
waitpoint, or ends when its timeout elapses.

```bash theme={null}
crewship routine waitpoints <subcommand> [flags]
```

The `pipeline` alias works in place of `routine`.

## Subcommands

| Command           | API behavior                                                                                      |
| ----------------- | ------------------------------------------------------------------------------------------------- |
| `list`            | `GET /api/v1/workspaces/{workspace}/pipelines/waitpoints`; lists pending waitpoints.              |
| `show <token>`    | Reads the pending list and prints full prompt and metadata for the exact token.                   |
| `approve <token>` | `POST /api/v1/workspaces/{workspace}/pipelines/waitpoints/{token}/approve` with `approved: true`. |
| `reject <token>`  | The same endpoint with `approved: false`.                                                         |

### `list`

```bash theme={null}
crewship routine waitpoints list
crewship routine waitpoints list --format json
```

Human output columns are `TOKEN RUN ID STEP KIND CREATED TIMEOUT PROMPT`.
The full token is shown because it is required by `approve` and `reject`; the
run ID is shortened only in the human table. An empty pending list prints
`No pending waitpoints.` in table mode and `[]` in JSON mode. `--json` is a
deprecated alias for `--format json`; YAML and NDJSON are also supported.

### `show <token>`

```bash theme={null}
crewship routine waitpoints show wpt_abc123
```

Prints token, run ID, step, kind, optional invoking crew, created/timeout
timestamps, optional callback URL, and the full prompt. A missing, expired, or
already-decided token returns a not-found error. If the server supplied a
callback URL, the command also prints the unauthenticated external completion
example; treat that URL as a secret.

### `approve` and `reject`

```bash theme={null}
crewship routine waitpoints approve wpt_abc123 --comment "LGTM"
crewship routine waitpoints reject wpt_abc123 --comment "Needs revision"
```

Both commands accept `--comment` (default empty). The comment is forwarded to
the parked run as the wait step's output; rejection additionally carries the
rejection signal. Human output is `Approved waitpoint <short-token>.` or
`Rejected waitpoint <short-token>.` Decisions require authentication and a
workspace and return the server's Problem Details error for an invalid or
already-resolved token.