Resume a paused or failed workflow run on the wise engine, or answer the gate of a gated one, then follow its events in this conversation. Invoked as `/wise-workflow-resume` (bare alias) or `/wise:wise-workflow-resume` (canonical). Use when the user says "resume the workflow", "continue the paused run", "pick up the run", "resume run <ulid>", or types `/wise-workflow-resume`.
Installs into .claude/skills of the current project.
Are you the author of Wise Workflow Resume?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/e1024kb-wise-workflow-resume)
---
name: wise-workflow-resume
description: >-
Resume a paused or failed workflow run on the wise engine, or answer
the gate of a gated one, then follow its events in this conversation.
Invoked as `/wise-workflow-resume` (bare alias) or
`/wise:wise-workflow-resume` (canonical). Use when the user says
"resume the workflow", "continue the paused run", "pick up the run",
"resume run <ulid>", or types `/wise-workflow-resume`.
argument-hint: "[<run-ulid>]"
allowed-tools: Read, Write, Skill, AskUserQuestion, TodoWrite, Task, Agent, TeamCreate, TeamDelete, SendMessage, Monitor, TaskCreate, TaskList, TaskGet, TaskUpdate, TaskOutput, TaskStop, Bash(bash:*), Bash(python3:*), Bash(test:*)
---
# /wise-workflow-resume
Before executing, follow [model fallback](../../references/workflow-host-control.md#model-fallback)
for unavailable models or delegation routes, including in autonomous procedures.
At every skill start, identify your main/child role and the current client
and GUI/TUI question tools, then read and follow the
[question lifecycle](../../references/workflow-host-control.md#keep-asynchronous-questions-open).
Keep asynchronous prompts open until answered; this rule does not authorize
questions in autonomous or otherwise prompt-free procedures.
First read [host control](../../references/workflow-host-control.md). Resolve the
loaded installation, set `WISE_HOST` to this conductor and `WISE_PLUGIN_ROOT`
to that installation. Use its managed launcher for shell commands. Follow the
reference's diagnostics and explicit-answer fallback when MCP or a native picker
is unavailable. Conductor host and child provider are independent.
Tools come from the managed `wise-engine` MCP server. Errors return
`{"error":{code,message,...}}`. Diagnose `DAEMON_UNAVAILABLE` through host
control before retrying. For `AUTH_REQUIRED`, print the selected provider's
`login_cmd` verbatim and stop.
## 1. Pick the run
`wise_status` (no id) lists every run. Resumable: `paused`, `failed`.
`gated` runs are answered, not resumed. `running` runs are only
followed.
- Argument given: `wise_status {run_id}`. On `RUN_NOT_FOUND`, use
`python3 "${WISE_PLUGIN_ROOT}/scripts/wise-helpers.py" runs-root`
to locate history. A `state.yaml` entry is a legacy v1 run: report
`Legacy run cannot resume. Import its definition, review it, then
start a new v2 run.` Keep its files unchanged. Otherwise print
`No run <run_id>.` and stop.
- No argument: use the host control reference's explicit-answer route over the `paused`, `failed`, `gated`
runs (label `run_id`; description `<workflow>, <status>, last
activity <last_activity_at>, <cwd>`) plus Abort. None: print
`No resumable runs. /wise-workflow-status lists all runs.` and stop.
## 2. Resume
Inspect the step statuses before resuming a failed run. Resume resets steps left
`running` to `pending`; it does not reset a step already marked `failed` and is
not a retry-failed-step command. If no runnable pending work remains, the run
fails again. Explain that limit, review completed side effects, and use a new
run after correcting the cause when a failed step needs another attempt.
- `paused` / `failed`: `wise_resume {run_id}` returns `{run_id,
status}`. Print `Resuming <run_id> (<workflow>).`
- `gated`: show `gate.message`, collect an explicit answer with `gate.options`
(free text when `allow_text`; beyond the host's option cap apply the
[long option list rule](../../references/workflow-host-control.md#long-option-lists)),
`wise_answer {run_id, gate_id, value}`.
- `completed` / `cancelled`: say the run is terminal and stop.
## 3. Follow
`wise_wait {run_id, after: 0, timeout_ms: 0}` returns the history at
once: print one line, `<n> steps done, <m> pending`, not the replay,
and set `after` to the last `seq`. Then run the wait loop and final
report exactly as `/wise-workflow-run` §4 and §5 (one line per event,
no raw step output, gates through the host control reference's explicit-answer
route and `wise_answer`,
`done: true` ends the loop).
## Rules
- Never re-run finished steps; the engine resumes only in-flight and
pending work.
- Do not invoke other wise skills.