Skip to content

Workflows

Commands

8 slash commands for common workflows. Each one delegates to a specialized agent through the task tool, or runs directly in the primary session.

$ /arch

Create an Architecture Decision Record (ADR) for a design choice

@software-architectsubtasksource
Prompt

Create an Architecture Decision Record for: $ARGUMENTS

!git log --oneline -5

Follow the standard ADR format:

# ADR-NNN: Title

## Status
[Proposed | Accepted | Deprecated | Superseded by ADR-NNN]

## Context
Forces, constraints, and rationale that require a decision.
Include the problem statement and any evaluated alternatives.

## Decision
The chosen approach — what and why. Reference specific patterns, libraries, or conventions.

## Consequences
Trade-offs, migration effort, operational burden.

Consider:

  • System context and constraints
  • Evaluated alternatives with pros/cons
  • Impact on performance, security, maintainability
  • Migration path if replacing an existing solution
$ /audit

Security audit of a file, directory, or the whole project

@security-auditorsubtasksource
Prompt

Perform a security audit on $ARGUMENTS.

!git ls-files

Analyze the codebase for:

  • Injection vulnerabilities (SQLi, XSS, command injection, SSTI)
  • Authentication and authorization flaws
  • Insecure direct object references (IDOR)
  • Sensitive data exposure (hardcoded secrets, keys, tokens)
  • Dependency vulnerabilities
  • Misconfigurations (CORS, CSP, headers, permissions)
  • Insecure deserialization
  • Use of vulnerable or deprecated packages

For each finding, provide:

  • File and line reference
  • Impact assessment (critical/high/medium/low)
  • Remediation steps with code example
  • CWE/CVE reference where applicable
$ /deploy

Prepare a release with changelog, version bump, and tag

@release-managersubtasksource
Prompt

Prepare a release for $ARGUMENTS.

!git log --oneline $(git describe --tags --abbrev=0 2>/dev/null || echo HEAD~20)..HEAD

Analyze the recent commits and:

  1. Determine the next version number (semver: major/minor/patch)
  2. Generate a changelog with clear sections:
    • Features
    • Bug Fixes
    • Performance Improvements
    • Breaking Changes
    • Deprecations
  3. Suggest the git tag command
  4. Check if any migration steps or release notes are needed
  5. Verify version references in:
    • package.json, Cargo.toml, pyproject.toml, or similar
    • Any changelog files
    • Documentation version references
$ /design

Review UI components, design tokens, and CSS architecture

@css-ui-specialistsubtasksource
Prompt

Review the design system and UI implementation in $ARGUMENTS.

!find $ARGUMENTS -type f \( -name '*.css' -o -name '*.tsx' -o -name '*.jsx' -o -name '*.vue' -o -name '*.scss' -o -name '*.module.css' \) 2>/dev/null | head -20

Evaluate:

  • Design tokens: Are colors, spacing, typography, and shadows using CSS custom properties or a token system?
  • Responsiveness: Are layouts responsive without horizontal overflow or magic breakpoints?
  • Accessibility: Color contrast (4.5:1 WCAG AA), focus indicators, aria attributes, semantic HTML
  • Performance: Unused CSS, render-blocking resources, layout shifts
  • Consistency: Repeated patterns that should be unified
  • Dark mode: Are there hardcoded light colors that break in dark mode?
  • Animation: Respects prefers-reduced-motion, purposeful not decorative
$ /docs

Generate or update documentation for the specified module or component

@docs-writersubtasksource
Prompt

Generate comprehensive documentation for $ARGUMENTS.

!find $ARGUMENTS -type f | head -30 2>/dev/null || echo "checking project structure"

Include:

  • Overview and purpose
  • Public API surface (functions, classes, exports)
  • Usage examples
  • Configuration options
  • Dependencies and requirements
  • Edge cases and error handling
  • Related files and references

Use the project's existing documentation style as reference: !find . -name '*.md' -not -path './node_modules/*' | head -10

Output the documentation in markdown format, ready to save to a file.

$ /refactor

Analyze and plan refactoring for a file, module, or pattern

@refactor-agentsubtasksource
Prompt

Analyze $ARGUMENTS for refactoring opportunities.

!wc -l $ARGUMENTS 2>/dev/null || echo "analyzing structure"

Evaluate:

  • Complexity: Functions/modules with high cyclomatic complexity
  • Duplication: Repeated code patterns that could be extracted
  • Coupling: Tightly coupled components that should be decoupled
  • Naming: Unclear or inconsistent naming conventions
  • Structure: Files/modules that violate single responsibility
  • Dead code: Unused functions, imports, or parameters
  • Type safety: Missing or incorrect type annotations

For each finding, provide:

  • Current code snippet
  • Proposed refactored version
  • Rationale for the change
  • Estimated effort (small/medium/large)

Do NOT make changes directly — only produce the analysis and recommendations.

$ /review

Review uncommitted changes for quality, bugs, and best practices

@code-reviewersubtasksource
Prompt

Review the following uncommitted changes for code quality, potential bugs, performance issues, and adherence to project conventions.

!git diff --stat !git diff

Focus on:

  • Logic errors and edge cases
  • Security vulnerabilities
  • Performance bottlenecks
  • Code style and consistency with the rest of the codebase
  • Missing tests or error handling

Provide a summary of findings with severity levels (critical/major/minor) and concrete suggestions for each issue.

$ /test

Run tests, analyze failures, and suggest fixes

@test-writersubtasksource
Prompt

Run the test suite and analyze the results.

!npm test 2>&1 || true

!npx vitest run 2>&1 || true

!go test ./... 2>&1 || true

!cargo test 2>&1 || true

Based on the test output:

  • Identify all failing tests and their error messages
  • Analyze the root cause of each failure
  • Suggest specific fixes with code examples
  • Check for missing test coverage in critical paths
  • Recommend additional test cases for edge conditions

If no test framework is detected, analyze the project structure and recommend an appropriate testing setup.