sdlc

name	sdlc
description	SDLC workflow automation that implements features using TDD skill, runs test-runner skill, and prepares commit commands for user review. Use when user provides a spec/plan for a new feature. Workflow: (1) invokes test-driven-development skill with spec, (2) invokes test-runner skill, (3) prepares git commands for user to execute manually.

SDLC - Software Development Lifecycle Automation

Overview

This skill automates the software development lifecycle workflow for implementing features:

Workflow: TDD Skill → Test-Runner Skill → Identify Files → Prepare Commit Commands

1. Invoke test-driven-development skill with the spec/plan
2. Invoke test-runner skill (handles lint, type checking, and all tests)
3. Identify all modified files related to the user's request
4. Prepare commit commands using /commit (saves to tmp/git-commands.txt for user to copy/paste)

⚠️ NOTE: This skill does NOT execute git commits or pushes. It prepares the commands for user review and manual execution.

When This Skill Is Invoked

Immediately execute the SDLC workflow. DO NOT ask questions first.

Execute these steps NOW:

1. Invoke the test-driven-development skill with the spec/plan
2. Invoke the test-runner skill
3. Identify all modified files related to the user's request
4. Prepare commit commands with /commit (user executes manually)

DO NOT:

• ❌ Ask "Would you like me to run the workflow?"
• ❌ Wait for user confirmation before starting
• ❌ Run just commands directly (let test-runner skill handle them)
• ❌ Execute git commit or push commands directly

JUST RUN IT. The user invoked this skill to complete the SDLC workflow.

Workflow

Step 0: DRY Check - Can We Reuse Existing Patterns?

BEFORE implementing, run semantic search to find similar implementations:

docker exec arsenal-semantic-search-cli code-search find "<what the spec does>"

Ask yourself:

1. Does similar functionality already exist?
2. Could we modify the spec slightly to reuse existing infrastructure?
3. Would extending existing code be simpler than building new?

If reuse is possible, STOP and propose spec modifications to the user:

⚠️ SPEC SIMPLIFICATION OPPORTUNITY

I found existing infrastructure that does something similar:
- [existing pattern/file]

Instead of implementing the spec as-is, we could:
- Option A: Extend [existing thing] with minor changes
- Option B: Adjust the spec to use [existing pattern]

This would reduce complexity by [benefit].

Should I adjust the approach?

Example: If spec says "create new YAML DSL for cron conditions" but you find group_message_intervention_conditions_yaml already handles similar logic, propose reusing/extending that instead.

Only proceed to Step 1 after confirming no reuse opportunity exists.

Step 1: Invoke TDD Skill

Invoke the test-driven-development skill with the spec/plan:

Use the test-driven-development skill to implement '/path/to/spec.md'

Or with inline spec:

Use the test-driven-development skill to implement: [description]

The TDD skill handles all implementation details (tests first, watch fail, minimal code to pass, refactor).

Step 2: Invoke Test-Runner Skill

Invoke the test-runner skill.

The test-runner skill handles:

• Lint and auto-fix (ruff)
• Type checking (mypy)
• Running all appropriate tests
• Ensuring code quality before commit

Critical: Do NOT proceed to prepare commits if test-runner reports failures.

Step 3: Identify Modified Files

After test-runner completes successfully:

• Run git status to see all modified files
• Identify which files are related to the user's request
• Include both:
  • Files modified by TDD skill (production code and test files)
  • Files that were auto-fixed by the linter

Step 4: Prepare Commit Commands

Run /commit with all modified files related to the user's request:

/commit api/src/services/user.py api/tests/unit/test_user.py

The /commit command:

1. Analyzes the changes
2. Creates appropriate conventional commit message(s)
3. Saves git commands to tmp/git-commands.txt as one-liners (does NOT execute them)
4. Tells user to run cat tmp/git-commands.txt to copy/paste

Step 5: User Executes Manually

After the workflow completes, inform the user:

✅ SDLC workflow complete!

Git commands saved to: tmp/git-commands.txt

To commit and push:
1. View commands: cat tmp/git-commands.txt
2. Copy/paste each line to execute
3. Push: git push origin <branch-name>

Usage Examples

Example 1: Implement from spec file

Use the SDLC skill to implement spec/2024-12-15-user-auth/01-authentication.md

Runs: TDD skill → test-runner skill → prepare commit commands

Example 2: Implement from inline description

Use the SDLC skill to add user authentication to the API

Runs: TDD skill → test-runner skill → prepare commit commands

Notes

• TDD First: Always invokes test-driven-development skill first
• Quality Gates: Always invokes test-runner skill before preparing commits
• Delegates to Skills: Does NOT run just commands directly — lets skills handle details
• Smart Commits: Only includes files related to the user's request
• Non-Executing: Git commands are saved to tmp/git-commands.txt for user to copy/paste
• No Questions: Executes immediately when invoked (don't ask for confirmation)
• No Broken Code: Never prepares commits if test-runner reports failures