Subcommand	Description
`start [target...]`	Start remote monitoring (tmux session)
`stop [target...]`	Stop remote monitoring
`status`	Show monitoring status
`auto`	Start auto-discovery mode
`auto stop`	Stop auto mode
`auto status`	Show auto mode status

Target Specification

<work-id>...: One or more work IDs (variadic)
--all: All works with GitHub source
<pattern>: Wildcard (e.g., FEAT-*, *-auth, FIX-???-*)
Omitted: Use active_work from state.json

Processing

1. Resolve Targets

From state.json .works, filter by target type:

--all: works with source.type == "github"
pattern: glob match against work IDs
multiple: verify each exists
single/omitted: use specified or active work

2. start

For each target work:

Get GitHub issue number from state.json source
Launch tmux session wf-remote-<work-id> running the polling daemon
Update state.json with remote: { enabled: true, source_issue, status: "waiting_approval", tmux_session }

3. stop

Kill tmux sessions for target works. Update state.json remote.enabled = false.

4. status

Display for each monitored work: issue number, status, tmux session state, last check time.

Remote Commands (Issue Comments)

Command	Description
`/approve` or `/next`	Execute next workflow step
`/pause`	Pause monitoring
`/stop`	Stop monitoring completely

State JSON Extension

{
  "works": {
    "<work-id>": {
      "remote": {
        "enabled": true,
        "source_issue": 123,
        "poll_interval": 60,
        "last_check": "<ISO8601>",
        "status": "waiting_approval|executing|paused|stopped",
        "tmux_session": "wf-remote-<work-id>"
      }
    }
  }
}

Security

Rule	Description
Collaborator-only	Only `admin`/`write`/`maintain` permission comments processed
Step limit	Max 10 steps per session
Command whitelist	Only `/approve`, `/next`, `/pause`, `/stop`
Execution scope	Only `/wf0-nextstep` executed

Auto Mode

Auto mode automatically discovers GitHub Issues with a specific label and processes them without manual intervention. It also supports revision mode for re-processing issues after PR review feedback.

Usage

/wf0-remote auto              # Start auto-discovery mode
/wf0-remote auto stop         # Stop auto mode
/wf0-remote auto status       # Show current status
/wf0-remote auto --max 3      # Process maximum 3 issues
/wf0-remote auto --dry-run    # List issues without executing
/wf0-remote auto --once       # Process one issue and exit

Options

Option	Description
`--max <N>`	Maximum issues to process (default: 5)
`--cooldown <MIN>`	Minutes between issues (default: 5)
`--dry-run`	Query issues without executing
`--once`	Process once and exit

Configuration

Add to .wf/config.json:

{
  "auto": {
    "query": "auto-workflow",
    "exclude_labels": ["blocked", "wip"],
    "complete_label": "completed",
    "revision_label": "needs-revision",
    "max_issues": 5,
    "cooldown_minutes": 5
  }
}

Workflow (New Issues)

Query GitHub Issues with configured label (excluding completed)
For each issue (oldest first):
- Create feature branch from base
- Execute /wf1-kickoff #<issue>
- Run /wf0-nextstep loop until completion
- Push changes and add completed label
On failure: post error comment and skip to next issue
Cooldown between issues to prevent rate limiting

Revision Mode

When a PR needs changes after review, the revision workflow automatically processes feedback:

Trigger

Add needs-revision label to an issue that already has completed label.

Revision Workflow

Auto-daemon detects issue with both completed and needs-revision labels
Revisions are prioritized over new issues
Existing workspace is restored via /wf0-restore
/wf1-kickoff revise incorporates:
- PR review comments
- Issue body updates
- New Issue comments
Full workflow re-executes (wf2 → ... → wf6 → wf7)
Additional commits are pushed to existing PR
needs-revision label is removed (keeping completed)

Revision Flow Diagram

[Human] Reviews PR, adds feedback comments
    ↓
[Human] Adds `needs-revision` label to Issue/PR
    ↓
[auto-daemon] Detects: completed + needs-revision
    ↓
[auto-daemon] wf0-restore (existing work-id)
    ↓
[auto-daemon] wf1-kickoff revise (PR/Issue feedback)
    ↓
[auto-daemon] wf0-nextstep loop (wf2 → ... → wf6)
    ↓
[auto-daemon] wf7-pr update (additional commits)
    ↓
[auto-daemon] Remove `needs-revision` label

State File

Auto mode maintains state in .wf/auto.json:

{
  "enabled": true,
  "session_start": "2026-01-30T10:00:00Z",
  "processed_count": 2,
  "current_issue": 456,
  "is_revision": false,
  "tmux_session": "wf-auto"
}

Field	Type	Description
`enabled`	boolean	Auto mode is running
`session_start`	string	Session start time (ISO8601)
`processed_count`	number	Issues processed this session
`current_issue`	number	Currently processing issue
`is_revision`	boolean	Current issue is a revision
`tmux_session`	string	tmux session name

Notes

Requires gh CLI and tmux
See rules/remote-operation.md for detailed security rules

wf0-remote

/wf0-remote

Usage

Subcommands