SKILL.md

--- frontmatter

name: devloop-audit
description: Audit devloop against Claude Code updates to identify integration opportunities. Use after Claude Code releases, monthly maintenance, or when exploring new features.
whenToUse: After Claude Code updates, periodic maintenance, exploring new Claude Code features, checking if devloop is current
whenNotToUse: Regular development work, quick tasks, when not working on devloop itself
disable-model-invocation: true
context: fork
agent: Explore
allowed-tools:
  - Read
  - Grep
  - Glob
  - WebFetch
  - WebSearch
  - Write
  - AskUserQuestion
seeAlso:
  - skill: plan-management
    when: creating plans from audit findings

Devloop Audit

Audit devloop against Claude Code's evolving capabilities to identify integration opportunities, deprecated patterns, and areas for improvement.

When to Run

•After major Claude Code releases
•Monthly maintenance check
•When noticing new Claude Code features
•Before major devloop changes

Audit Process

Step 1: Gather Claude Code Information

Fetch the latest documentation and release notes:

•
Check Release Notes
- •Fetch changelog: https://code.claude.com/docs/en/changelog.md
- •Use WebFetch to get the latest entries
- •Focus on: new tools, hooks, skills, plugin features, breaking changes
•
Review Key Documentation
- •Skills: https://code.claude.com/docs/en/skills
- •Plugins: https://code.claude.com/docs/en/plugins
- •Hooks: https://code.claude.com/docs/en/hooks
- •Sub-agents: https://code.claude.com/docs/en/sub-agents
•
Check Built-in Tools
- •TaskCreate/TaskUpdate/TaskList (session-scoped tasks)
- •Skill tool (skill invocation)
- •Task tool (subagent spawning)

Step 2: Analyze Current devloop Features

Review devloop's current implementation:

bash

# List all devloop commands
ls plugins/devloop/commands/

# List all devloop agents
ls plugins/devloop/agents/

# List all devloop skills
ls plugins/devloop/skills/

Step 3: Compare and Identify Gaps

For each devloop feature, check:

Category	Questions
Overlap	Does native Claude Code now provide this?
Enhancement	Can we leverage new features to improve this?
Deprecation	Are we using patterns that are now outdated?
Missing	Are there new capabilities we should adopt?

Step 4: Generate Findings Report

Priority Ranking Logic:

Priority	Criteria
High	Native feature fully replaces devloop feature, OR security/correctness issue, OR blocking user workflows
Medium	Native feature partially overlaps, OR improves UX/performance, OR aligns with best practices
Low	Minor improvement, OR "nice to have", OR requires significant refactoring for small benefit

Effort Estimation:

Effort	Description
Low	< 1 day, localized change, no breaking changes
Medium	1-3 days, touches multiple files, may require migration
High	3+ days, architectural change, breaking changes likely

Create a findings report at .devloop/spikes/claude-code-audit-YYYY-MM-DD.md:

markdown

# Claude Code Audit Findings

**Date**: YYYY-MM-DD
**Claude Code Version**: X.Y.Z
**devloop Version**: X.Y.Z

## Summary

Brief overview of findings.

## Integration Opportunities

### High Priority

| Finding | Current State | Recommended Action |
|---------|--------------|-------------------|
| [Description] | [How devloop works now] | [What to change] |

### Medium Priority

...

### Low Priority

...

## Deprecated Patterns

| Pattern | Native Alternative | Migration Path |
|---------|-------------------|----------------|
| [Current approach] | [New approach] | [How to migrate] |

## New Features to Adopt

| Feature | Use Case | Implementation Notes |
|---------|----------|---------------------|
| [Feature name] | [How it helps devloop] | [How to implement] |

## No Action Needed

Features that are already aligned with Claude Code best practices.

## Next Steps

1. Create spike plans for high-priority items
2. Schedule medium-priority items for future sprints
3. Document low-priority items for consideration

Step 5: Generate Spike Plan (if needed)

If high-priority items are found, generate a spike plan:

markdown

# Spike: Claude Code Integration Opportunities

**Date**: YYYY-MM-DD
**Source**: Audit findings from [date]
**Focus**: High-priority integration opportunities

## Questions to Answer

1. [Question derived from top finding]
2. [Question derived from second finding]

## Investigation Tasks

- [ ] Task 1: [Specific investigation from findings]
- [ ] Task 2: [Specific investigation from findings]

## Success Criteria

- Clear recommendation for each integration opportunity
- Migration path documented for any deprecations
- Risk assessment for breaking changes

Save to .devloop/spikes/claude-code-integration-YYYY-MM-DD.md

Step 6: Offer Next Steps

After generating the report, ask the user:

yaml

AskUserQuestion:
  questions:
    - question: "Audit complete. What would you like to do?"
      header: "Action"
      multiSelect: false
      options:
        - label: "Create spike from findings"
          description: "Start /devloop:spike with top priority items"
        - label: "Create plan from findings"
          description: "Create implementation plan for all items"
        - label: "Review findings"
          description: "Just review the report for now"

Audit Checklist

Commands

• /devloop - State detection still relevant?
• /devloop:run - Autonomous execution aligned with native patterns?
• /devloop:spike - Exploration workflow current?
• /devloop:fresh - Context management vs native summarization?
• /devloop:ship - Git integration patterns current?

Skills

• Skill format using standard frontmatter?
• Skills discoverable by Claude when relevant?
• Skill content appropriate for invocation type?

Agents

• Agent definitions using standard format?
• Subagent patterns aligned with Task tool?
• Agent capabilities appropriate for tasks?

Hooks

• Hook events using current API?
• Hook patterns following best practices?

Task Management

• Plan persistence (markdown) vs session tasks (TaskCreate)?
• Appropriate use of each system?
• Clear documentation on when to use which?

State Management

• .devloop/ directory structure current?
• State files using appropriate formats?
• Session state vs persistent state clear?

Key Documentation Links

Start here for complete doc index:

•Full Index: https://code.claude.com/docs/llms.txt

Core documentation pages:

•Skills: https://code.claude.com/docs/en/skills
•Plugins: https://code.claude.com/docs/en/plugins
•Plugins Reference: https://code.claude.com/docs/en/plugins-reference
•Hooks: https://code.claude.com/docs/en/hooks
•Sub-agents: https://code.claude.com/docs/en/sub-agents
•Interactive Mode: https://code.claude.com/docs/en/interactive-mode
•Memory: https://code.claude.com/docs/en/memory
•Permissions/IAM: https://code.claude.com/docs/en/iam

Example Output

markdown

# Claude Code Audit Findings

**Date**: 2026-01-27
**Claude Code Version**: 2.1.x
**devloop Version**: 3.14.0

## Summary

Found 3 integration opportunities: 1 high, 1 medium, 1 low priority.
No deprecated patterns detected.

## Integration Opportunities

### High Priority

| Finding | Current State | Recommended Action | Effort |
|---------|--------------|-------------------|--------|
| TaskCreate/Update native integration | devloop uses plan.md only | Hybrid: plan.md for persistence + native Tasks for UI | Medium |

### Medium Priority

| Finding | Current State | Recommended Action | Effort |
|---------|--------------|-------------------|--------|
| Skill frontmatter alignment | Some skills missing new fields | Update frontmatter to include `context`, `agent` where applicable | Low |

### Low Priority

| Finding | Current State | Recommended Action | Effort |
|---------|--------------|-------------------|--------|
| Document /compact vs /fresh | Users confused about when to use which | Add guidance to CLAUDE.md | Low |

## Deprecated Patterns

None found.

## New Features to Adopt

| Feature | Use Case | Implementation Notes |
|---------|----------|---------------------|
| `context: fork` for skills | Run expensive operations in subagent | Consider for devloop-audit skill |

## Next Steps

1. Create Issue #20 for TaskCreate integration (high priority)
2. Update skill frontmatter in next maintenance pass
3. Add /compact vs /fresh guidance to docs

Output

This skill produces:

•A findings report at .devloop/spikes/claude-code-audit-YYYY-MM-DD.md
•Summary of integration opportunities
•Option to create spike/plan for follow-up work