mycelium-work

name	mycelium-work
description	Executes implementation tasks following strict TDD methodology with evidence-based verification. Use when user says "implement this", "work on [task]", "execute the plan", "build task [id]", or "run all tasks". Enforces RED→GREEN→REFACTOR cycle, shows actual test output, supports parallel execution via worktrees.
license	MIT
version	0.9.0
argument-hint	[task_id or 'all']
allowed-tools	["Skill","Read","Write","Edit","Bash","Glob","Grep","Task"]
metadata	{"author":"Jason Hsieh","category":"implementation","tags":["tdd","evidence-based","parallel-execution","git-worktrees"],"documentation":"https://github.com/jason-hchsieh/mycelium"}

Workflow Work

Execute implementation tasks following strict TDD methodology.

Your Task

1. Update session state - Write invocation_mode: "single" to .mycelium/state.json

2. Load execution skills:

   • Use Skill tool to load tdd (mandatory)
   • Use Skill tool to load verification (mandatory)

3. Parse arguments:

   • task_id: Specific task (e.g., "1.1")
   • all: Execute all unblocked tasks (parallel)
   • Default: Next available task

4. Load active plan:

   • Read current_track.plan_file from state.json to locate the active plan
   • Fall back to latest plan in .mycelium/plans/ if current_track is null (backward compat)
   • Read state.json for progress

5. Execute tasks - Follow TDD and verification skills which handle:

   • TDD cycle: RED → GREEN → REFACTOR (mandatory)
   • Evidence-based verification: Show actual test output
   • Incremental commits: After each task completion
   • Plan marker updates: [ ] → [~] → [x]
   • Parallel execution: Worktrees for independent tasks (automatic)
   • Solution capture: For novel problems (as needed)

6. After completion: Suggest /mycelium-review for code review

Skills Used

• tdd: Iron Law TDD - tests first, always (MANDATORY)
• verification: Evidence-based validation (MANDATORY)
• mycelium-capture: Capture learnings from novel problems (as needed)

Quick Example

/mycelium-work          # Execute next task
/mycelium-work 1.1      # Execute specific task
/mycelium-work all      # Execute all tasks (parallel)

Critical Rules

• NO CODE WITHOUT TESTS FIRST - TDD is non-negotiable
• SHOW EVIDENCE - All verification requires actual test output
• UPDATE MARKERS - Plans updated in real-time during execution
• SAVE STATE - Frequent saves enable resume on interruption

Parallel Execution

When executing "all" tasks:

• Independent tasks (blockedBy: []) run in parallel
• Each task gets its own worktree
• Automatic merge on completion
• Full test suite runs after merge

References

• .mycelium/ directory structure
• Session state docs
• Session state schema
• Plan frontmatter schema

Examples

Example 1: Execute Specific Task

Command: /mycelium-work 1.1

Workflow:

1. Marks task 1.1 as [~] in progress
2. Loads TDD skill - enforces test-first
3. Executes: Write test → Verify RED → Implement → Verify GREEN
4. Commits with evidence
5. Marks task as [x] complete

Result: Task 1.1 done with passing tests

Example 2: Execute All Tasks

Command: /mycelium-work all

Workflow:

1. Identifies unblocked tasks (blockedBy: [])
2. Creates worktrees for parallel execution
3. Spawns agents per task
4. Each agent follows TDD cycle
5. Merges on completion, runs full test suite

Result: All independent tasks complete in parallel

Troubleshooting

Error: "Baseline tests failing" Cause: Existing tests broken before new work Solution: Fix existing tests first, then proceed with new work

Error: "Tests still failing after 3 attempts" Cause: Implementation approach or test is incorrect Solution: Use /recovery skill for systematic debugging

Issue: "Cannot create worktree" Cause: Branch already has a worktree or git repository issue Solution: Check git worktree list, remove stale worktrees with git worktree remove