| name | mycelium-view |
| description | Previews complete workflow plan without execution (dry-run mode). Use when user says "show me the plan", "what would this look like", "preview [task]", "dry run", or wants to see task breakdown before committing. Supports --verbose to display workflow diagram. No implementation, tests, or reviews executed. |
| license | MIT |
| version | 0.9.0 |
| argument-hint | [task description] | [track_id] [--verbose] |
| allowed-tools | ["Skill","Read","Write","Bash","Glob","Grep"] |
| metadata | {"author":"Jason Hsieh","category":"planning","tags":["preview","dry-run","planning"],"documentation":"https://github.com/jason-hchsieh/mycelium"} |
Mycelium View (Dry-Run Preview)
Preview the complete workflow plan that would be generated for a task without executing any implementation, tests, or reviews.
Purpose
This skill provides a "dry-run" mode that:
- Shows exactly what work would be planned
- Displays the task breakdown and dependencies
- Estimates effort and complexity
- Identifies which agents/skills would be used
- Does NOT execute any implementation
Your Task
-
Parse arguments:
task description: The feature/fix/optimization to preview (generates a new plan)
track_id: An existing plan's track ID to view (e.g., auth_20260211)
--verbose: Show extended details with workflow diagram (optional)
- To distinguish: if the argument (after removing
--verbose flag) matches a track_id in session_state.plans[] or a plan file in .mycelium/plans/, treat it as an existing plan view. Otherwise, treat it as a new task description.
- Extract
--verbose flag if present and set verbose_mode boolean
-
Route by argument type:
View Existing Plan
If track_id was provided:
- Find the plan file: check
session_state.plans[] for the plan_file, or glob .mycelium/plans/*{track_id}*.md
- Read the plan file and parse its frontmatter and content
- Display the plan (same format as step 5 below, but from the existing file)
- Show plan status from
plans[] registry (in_progress, paused, completed, etc.)
- Suggest next action based on status:
paused → /mycelium-plan --switch {track_id} to resume
in_progress → /mycelium-work to continue implementation
completed → /mycelium-capture {track_id} to capture learnings
preview → /mycelium-go "{description}" to execute
Generate New Preview
If task description was provided:
-
Update session state - Write invocation_mode: "view" to .mycelium/state.json
-
Execute planning workflow (see mycelium-plan)
-
Generate plan - Follow the planning workflow:
- Clarify requirements if ambiguous
- Discover available capabilities
- Create detailed task breakdown
- Define test strategy
- Identify dependencies
-
Display preview - Show the complete plan. Format depends on verbose_mode:
Brief mode (default):
## Workflow Preview: [Feature Name]
### Overview
[Brief description of what would be built]
### Success Criteria
[Measurable outcomes that define "done"]
### Phases & Tasks
#### Phase 1: [Phase Name]
- Task 1.1: [Task title] (Complexity: S/M/L, Agent: general-purpose)
- Task 1.2: [Task title] (Complexity: S/M/L, Agent: bash)
#### Phase 2: [Phase Name]
- Task 2.1: [Task title] (blockedBy: [1.1], Complexity: M)
### Parallel Execution Plan
[Show which tasks can run in parallel]
### Estimated Timeline
- Total tasks: X
- Parallel-capable: Y
- Sequential-only: Z
### Test Strategy
[TDD approach for each phase]
### Git Strategy
- Branch: feature/[name]
- Worktrees: [if multiple features]
Verbose mode (--verbose flag):
## Workflow Preview: [Feature Name]
### Overview
[Brief description of what would be built]
### Success Criteria
[Measurable outcomes that define "done"]
### Workflow Diagram
[ASCII art diagram showing phase flow and task dependencies - see below]
### Phases & Tasks
#### Phase 1: [Phase Name]
##### Task 1.1: [Task title]
- **Complexity:** S (50-200 lines, 1-2 files, 30-120 min)
- **Agent:** general-purpose
- **Skills:** tdd, verification
- **Model:** sonnet
- **Blocked by:** None (or task IDs)
- **Blocks:** [1.2, 2.1] (or "None")
**Description:**
[Full task description from plan]
**Acceptance Criteria:**
- [ ] Criterion 1
- [ ] Criterion 2
**Test Plan:**
[How to verify this works]
---
##### Task 1.2: [Task title]
[Same detailed format...]
#### Phase 2: [Phase Name]
[Same detailed format for all tasks...]
### Parallel Execution Plan
[Show which tasks can run in parallel]
### Estimated Timeline
- Total tasks: X
- Parallel-capable: Y
- Sequential-only: Z
### Test Strategy
[TDD approach for each phase]
### Git Strategy
- Branch: feature/[name]
- Worktrees: [if multiple features]
-
Generate workflow diagram (if verbose_mode is true):
Create ASCII art diagram showing:
- Phase flow: Plan → Work → Review → Capture (horizontal boxes with arrows)
- Task layout: Tasks grouped by phase, with dependencies shown vertically
- Parallel execution: Tasks at same vertical level = can run in parallel
- Status indicators: [ ] pending, [~] in progress, [✓] completed
Example diagram:
Workflow Diagram:
┌─────────────────────────────────────────────────────────────────────┐
│ MYCELIUM WORKFLOW │
└─────────────────────────────────────────────────────────────────────┘
┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
│ PLAN │─────▶│ WORK │─────▶│ REVIEW │─────▶│ CAPTURE │
└──────────┘ └──────────┘ └──────────┘ └──────────┘
│ │ │ │
▼ ▼ ▼ ▼
Phase 1 Phase 2 Phase 3 Phase 4
─────── ─────── ─────── ───────
[ ] 1.1 [ ] 2.1 [ ] 3.1 [ ] 4.1
│ ┌┴─┐ │ │
▼ ▼ ▼ ▼ ▼
[ ] 1.2 [ ] 2.2 2.3 [ ] 3.2 [ ] 4.2
│ └┬─┘
▼ ▼
[ ] 1.3 [ ] 2.4
Legend:
─────▶ Sequential flow (phase to phase)
│ ▼ Dependency (task to task)
┌──┐ Parallel tasks (same vertical level)
[ ] Pending task
[~] In progress
[✓] Completed
Diagram generation algorithm:
- Parse plan to extract tasks, statuses, and
blockedBy dependencies
- Group tasks by phase (based on task ID prefix: 1.x → Phase 1, 2.x → Phase 2, etc.)
- Within each phase, calculate vertical levels:
- Level 0: Tasks with
blockedBy: [] (no dependencies, can start immediately)
- Level N: Tasks blocked only by tasks at level N-1
- Tasks at same level can execute in parallel
- Render using box-drawing characters:
─ │ ┌ ┐ └ ┘ ├ ┤ ┬ ┴ ┼ ▶ ▼
- Use task status from plan frontmatter if viewing existing plan, or
[ ] for new previews
-
Save preview - Write plan to .mycelium/plans/preview-YYYY-MM-DD-{track-id}.md with status: preview in frontmatter
-
Prompt next action:
✅ Workflow preview complete!
To execute this plan:
- Run full workflow: /mycelium-go "[task description]"
- Run phases individually: /mycelium-plan, /mycelium-work, etc.
To modify the plan:
- Edit: .mycelium/plans/preview-[timestamp].md
- Re-run: /mycelium-view "[modified description]"
Quick Examples
/mycelium-view "Add user authentication with JWT"
/mycelium-view auth_20260211
/mycelium-view "Add user authentication with JWT" --verbose
/mycelium-view --verbose auth_20260211
/mycelium-view "Fix memory leak in session handler"
/mycelium-view --verbose "Refactor API layer to use repository pattern"
Differences from /mycelium-go
| Feature | /mycelium-view | /mycelium-go |
|---|
| Creates plan | ✓ | ✓ |
| Executes implementation | ✗ | ✓ |
| Runs tests | ✗ | ✓ |
| Runs review | ✗ | ✓ |
| Captures learnings | ✗ | ✓ |
| Makes code changes | ✗ | ✓ |
| Git commits | ✗ | ✓ |
Use Cases
Use /mycelium-view when:
- Estimating effort before committing
- Understanding scope and complexity
- Getting approval for approach
- Learning about mycelium's planning process
- Checking if requirements are clear enough
Use /mycelium-go when:
- Ready to execute the full workflow
- Requirements are clear
- Want autonomous implementation
Important
- Read-only mode - No code changes, commits, or state mutations (except state.json and preview plan file)
- Clarification allowed - Will ask questions if requirements are ambiguous
- Fast execution - Only planning phase runs (~30-60 seconds)
- Repeatable - Safe to run multiple times with different descriptions
- Preview state - Plans marked with
status: preview won't interfere with active work
- Verbose mode - Use
--verbose flag for extended details, workflow diagram, and full task specifications
- Diagram compatibility - ASCII art diagram uses standard box-drawing characters compatible with all terminals
Session State
The session state is updated to indicate view mode:
{
"invocation_mode": "view",
"current_phase": "planning",
"preview_mode": true,
"last_preview": "YYYY-MM-DD-preview-{track-id}.md"
}
This prevents accidental execution if /mycelium-continue is called after a preview.
Output Format
The skill outputs:
- Terminal display - Formatted markdown preview (brief or verbose depending on
--verbose flag)
- File output -
.mycelium/plans/preview-YYYY-MM-DD-{track-id}.md
- Workflow diagram - ASCII art visualization (only in
--verbose mode)
- Next steps - Clear instructions on how to proceed
Skills Used
- mycelium-plan: Requirements gathering and task decomposition (same workflow, preview mode)
Notes
- Preview plans are saved separately from execution plans
- The preview doesn't consume git commits or create branches
- Safe to use in dirty working directories
- Can be used as documentation for stakeholders
References