tapania

5d-build

"Execute implementation tasks with quality and awareness. Use when: (1) After TASKS phase in 5D-SDD workflow, (2) Tasks are defined and sequenced, (3) User is ready to write code, (4) User asks to 'implement,' 'build,' or 'code' a task. This phase produces working code while maintaining connection to spec intent."

tapania 3 Updated 4mo ago
GitHub

Install

npx skillscat add tapania/5d-skills/5d-build

Install via the SkillsCat registry.

SKILL.md

BUILD Phase

Execute tasks while maintaining spec alignment and code quality.

Build Principles

Spec Fidelity

Before writing code, re-read the relevant spec section. After writing code, verify it matches spec.

Common drift patterns:

  • Adding features not in spec ("while I'm here...")
  • Simplifying away spec requirements
  • Changing interfaces without updating spec

If drift is intentional: update spec first, then code.

Level 3 Awareness

The implementation is one solution, not the only solution. Stay open to:

  • Better approaches discovered during coding
  • Spec errors revealed by implementation reality
  • Edge cases the spec didn't anticipate

Flag these for VERIFY phase rather than silently changing.

Domain Connection (Width)

While building, ask: "Does this still serve the business/user intent?"

Technical correctness ≠ solving the right problem.

Quadrant Awareness

During implementation, check all perspectives:

  • Individual Outer: Is the code artifact correct?
  • Individual Inner: Do I understand what I'm building?
  • Collective Outer: Does this integrate with existing systems?
  • Collective Inner: Am I aligned with team/stakeholder expectations?

Skill Dependencies (Height)

If blocked:

  • Is this a code problem or a knowledge problem?
  • What adjacent skill would unblock this?
  • Should I pause to learn, or flag for help?

Time Awareness

  • Am I building on existing patterns (transcend and include)?
  • Am I replacing something? Is deprecation handled?
  • Am I creating future technical debt?

Identity Trap

Watch for defensive coding:

  • Adding complexity to avoid admitting spec issues
  • Protecting assumptions rather than solving problems
  • "It works" vs "it's right"

Build Process

For each task:

  1. Read the task description and done criteria
  2. Read the relevant spec section
  3. Implement the code
  4. Self-verify against done criteria
  5. Note any spec drift, discoveries, or concerns

Quality Checks During Build

  • Does it compile/run?
  • Does it match the spec interface?
  • Are error cases handled?
  • Is it testable?
  • Would you be comfortable debugging this at 3am?

Output Per Task

## Task [N] Complete

**Status:** done / blocked / needs-spec-update

**Implementation notes:**
[Brief description of approach taken]

**Spec drift:**
[Any deviations from spec - intentional or discovered]

**Discoveries:**
[Edge cases, issues, or insights found during implementation]

**Ready for verify:** [yes/no]

When to Pause

Stop building and escalate if:

  • Spec is ambiguous or contradictory
  • Implementation reveals spec error
  • Complexity far exceeds estimate
  • Blocked by unresolved dependency

Return to appropriate earlier phase rather than guessing.

Exit Criteria

Proceed to VERIFY when:

  • All tasks in current batch are complete
  • Code runs without errors
  • Self-verification passes
  • Discoveries are documented