intent-build-now

intent-build-now

Start implementation from Intent. Validates Intent completeness, then either delegates to TaskSwarm (if available) or executes TDD phases directly. Use when you have an Intent ready and want to start building.

5stars
0forks
Updated 2/3/2026
SKILL.md
readonlyread-only
name
intent-build-now
description

Start implementation from Intent. Validates Intent completeness, then either delegates to TaskSwarm (if available) or executes TDD phases directly. Use when you have an Intent ready and want to start building.

Intent Build Now

Start building from Intent. Validates completeness, then chooses execution path.

Workflow

/intent-build-now [path]
       ↓
Locate Intent + plan.md
       ↓
Validate completeness ──→ Incomplete? ──→ Show gaps, suggest /intent-plan
       ↓
Check for TaskSwarm
       │
       ├── Has TaskSwarm + TASK.yaml ──→ Delegate to /swarm run
       │
       └── No TaskSwarm ──→ Execute TDD phases directly

Step 1: Locate Files

Search for Intent files in order:

  1. User-specified path
  2. intent/{name}/ directories
  3. Current directory

Required files:

  • INTENT.md - The intent document
  • plan.md - Execution plan with test matrix

Optional:

  • TASK.yaml - TaskSwarm status file

If plan.md missing:

plan.md not found.

Run /intent-plan first to generate the execution plan.

Step 2: Validate Intent Completeness

Required Sections in INTENT.md

Section Purpose How to Fix
Responsibilities What it does / doesn't do Use /intent-interview to clarify scope
Structure ASCII diagram of components Add ## Structure with ASCII diagram
API Function signatures Define key interfaces in ## API

Required in plan.md

Check Criteria
Phases Has ## Phase 0: or more phases
Tests Each phase has 6 test categories
E2E Gate Each phase has ### E2E Gate
Checkboxes Uses - [ ] format

If Not Ready

## Intent Validation: NEEDS WORK

Missing elements:

1. **plan.md missing E2E Gate for Phase 1**
   - Add `### E2E Gate` with verification script

2. **plan.md Phase 0 missing Data Leak tests**
   - Add `#### Data Leak` section with test cases

Run /intent-plan to regenerate, or fix manually.

Step 3: Choose Execution Path

Path A: TaskSwarm Available

Detect TaskSwarm by checking:

  1. TASK.yaml exists in intent directory
  2. TaskSwarm plugin is installed (check for /swarm skill)

If both conditions met:

## Intent Validation: PASSED ✓

TaskSwarm detected. Delegating execution...

Your Intent is ready. Run:

    /swarm run {task_name}

Or for continuous execution:

    /swarm run-all

TaskSwarm will:
- Claim the task atomically
- Execute phases with TDD discipline
- Update checkboxes in plan.md
- Commit after each phase
- Push to remote

Do NOT execute phases yourself when TaskSwarm is available.

Path B: No TaskSwarm (Direct Execution)

If TaskSwarm not available:

## Intent Validation: PASSED ✓

No TaskSwarm detected. Executing directly...

Starting Phase 0: {phase_name}

Then execute TDD loop for each phase:

For each Phase:
  1. Read ### Tests section
  2. For each unchecked test `- [ ]`:
     a. Write test code (if not exists)
     b. Run test → expect failure (red)
     c. Write implementation
     d. Run test → expect pass (green)
     e. Update plan.md: `- [ ]` → `- [x]`
  3. Run E2E Gate script
  4. Git commit: "feat({scope}): Phase {n} - {phase_name}"
  5. Continue to next Phase

Step 4: Completion

After all phases complete:

## Build Complete ✓

All phases executed:
- Phase 0: {name} ✓
- Phase 1: {name} ✓

Next steps:
- Run /intent-sync to update Intent with confirmed details
- Run /intent-check to verify consistency

Integration

/intent-interview     # Create Intent from scratch
    ↓
/intent-review        # Section-by-section approval
    ↓
/intent-plan          # Generate plan.md + TASK.yaml
    ↓
/intent-build-now     # THIS SKILL
    ↓
┌───────────────────────────────────────┐
│  Has TaskSwarm?                       │
│    Yes → /swarm run (delegate)        │
│    No  → Execute TDD directly         │
└───────────────────────────────────────┘
    ↓
/intent-sync          # Sync confirmed details back

TDD Execution Standards (Direct Mode)

When executing without TaskSwarm, follow strict TDD:

Test Categories (All 6 Required)

  1. Happy Path - Normal expected usage
  2. Bad Path - Invalid inputs, error conditions
  3. Edge Cases - Boundary conditions
  4. Security - Vulnerability prevention
  5. Data Leak - Information exposure prevention
  6. Data Damage - Data integrity protection

TDD Discipline

  • Tests First: Write ALL tests before implementation
  • Red-Green-Refactor: Verify test fails, implement, verify pass
  • No Skip: Every checkbox must be completed
  • Commit Per Phase: One commit after each phase passes

E2E Gate Enforcement

Each phase's E2E Gate script MUST pass before proceeding:

# Run the E2E Gate script from plan.md
# Example: pnpm test -- --grep "Phase 0"

# If fails: Stop and fix
# If passes: Continue to next phase

Tips

  1. Prefer TaskSwarm - It handles concurrency, heartbeat, and recovery
  2. Don't force incomplete Intents - Missing plan.md means not ready
  3. Respect checkboxes - They're the source of truth for progress
  4. Commit frequently - One commit per phase, not one mega-commit

You Might Also Like

Related Skills

coding-agent

coding-agent

179Kdev-codegen

Run Codex CLI, Claude Code, OpenCode, or Pi Coding Agent via background process for programmatic control.

openclaw avataropenclaw
Get
add-uint-support

add-uint-support

97Kdev-codegen

Add unsigned integer (uint) type support to PyTorch operators by updating AT_DISPATCH macros. Use when adding support for uint16, uint32, uint64 types to operators, kernels, or when user mentions enabling unsigned types, barebones unsigned types, or uint support.

pytorch avatarpytorch
Get
at-dispatch-v2

at-dispatch-v2

97Kdev-codegen

Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.

pytorch avatarpytorch
Get
skill-writer

skill-writer

97Kdev-codegen

Guide users through creating Agent Skills for Claude Code. Use when the user wants to create, write, author, or design a new Skill, or needs help with SKILL.md files, frontmatter, or skill structure.

pytorch avatarpytorch
Get

Implements JavaScript classes in C++ using JavaScriptCore. Use when creating new JS classes with C++ bindings, prototypes, or constructors.

oven-sh avataroven-sh
Get

Creates JavaScript classes using Bun's Zig bindings generator (.classes.ts). Use when implementing new JS APIs in Zig with JSC integration.

oven-sh avataroven-sh
Get