- New `/xcode-test` command for building and testing iOS apps
- Uses XcodeBuildMCP to discover projects, build, install, launch
- Takes screenshots and captures console logs
- Supports human verification for Sign in with Apple, push, IAP
- Checks for XcodeBuildMCP installation before proceeding
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- New `/playwright-test` command for end-to-end browser tests on PR-affected pages
- Uses Playwright MCP to navigate, snapshot, check console errors
- Supports human-in-the-loop for OAuth/email/payment flows
- Creates P1 todos for failures and retries until passing
- Added Section 7 to `/workflows:review` - optional Playwright testing as subagent
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Router-pattern skill with sectioned references:
- controllers.md: REST mapping, concerns, Turbo, API patterns
- models.md: Concerns, state records, callbacks, scopes
- frontend.md: Turbo, Stimulus, CSS architecture
- architecture.md: Routing, auth, jobs, caching, config
- gems.md: What they use vs avoid, decision framework
Based on analysis of Fizzy (Campfire) codebase.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
New review agents for validating database migrations and risky data deployments:
- data-migration-expert: Validates ID mappings match production reality, checks
for swapped values, verifies rollback safety, provides SQL verification snippets
- deployment-verification-agent: Produces Go/No-Go deployment checklists with
pre/post-deploy SQL queries, data invariants, rollback procedures, monitoring
Updated /workflows:review to conditionally run these agents when PRs contain
database migrations (db/migrate/*.rb), data backfills, or ID/enum mappings.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Make description more specific to SKILL.md fixes, less likely to
trigger when running /workflows:compound
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixes#41
The /prime command was a personal setup command that shouldn't have been
included in the distributed plugin. It caused confusion when users had
hooks that auto-invoked it on session start.
Users who want a prime command can add it to their personal ~/.claude/commands/
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Fixes#42
Workflow commands now use `workflows:` prefix to avoid collisions with
Claude Code's built-in commands:
- `/workflows:plan` (was `/plan`)
- `/workflows:review` (was `/review`)
- `/workflows:work` (was `/work`)
- `/workflows:compound` (was `/compound`)
To upgrade: Update to latest plugin version. If you see duplicate `/plan`
commands, the new namespaced commands avoid this collision.
Also fixed in this release:
- Added missing `name:` field to heal-skill.md and create-agent-skill.md
- Rewrote corrupted prime.md
- Shortened Playwright MCP alias from `playwright` to `pw`
- Removed deprecated codify.md
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add agent-native-reviewer agent to verify features are agent-accessible
- Add agent-native-architecture skill for prompt-native design patterns
- Add agent-native-reviewer to /review command parallel agents
- Move agent-native skill to correct plugin folder
- Update component counts (25 agents, 12 skills)
- Include mermaid dark mode fix from PR #45🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Made capturing screenshots REQUIRED for any UI changes
- Updated to use imgup skill with pixhost as default host
- Clarified what to capture: new screens, before/after, Figma matches
- Removed outdated curl/0x0.st reference
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
New skill teaching prompt-native development patterns:
- Features defined in prompts, not code
- Tools as primitives that enable capability
- "Whatever the user can do, the agent can do"
- Self-modification patterns (advanced tier)
- Refactoring guide for existing codebases
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added "Open plan in editor" as the first option in post-generation menu
- Opens plan file in user's default editor for review before next steps
- Reordered options: open → review → work → work remote → create issue → simplify
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added step to capture before/after screenshots for UI changes
- Use Playwright MCP tools for screenshots
- Upload via 0x0.st (imgup skill)
- Updated PR template with before/after table format
- Updated quality checklist
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Reorder options: Review first, Work locally second, Work on remote third
- Add "Work on remote" option with `&` for background execution on Claude Code web
- Remove separate "Rework" option since "Other" handles custom changes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
BREAKING: Plugin renamed from compounding-engineering to compound-engineering.
Users will need to reinstall with the new name:
claude /plugin install compound-engineering
Changes:
- Renamed plugin directory and all references
- Updated documentation counts (24 agents, 19 commands)
- Added julik-frontend-races-reviewer to docs
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add critical documentation about Gemini returning JPEG by default
- Explain that using .jpg extension avoids "Image does not match media type" errors
- Provide PNG conversion example when needed
- Include format verification command
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Changed default model to gemini-2.0-flash-exp-image-generation
- Removed regular Nano Banana model references
- Added explicit options for aspect ratio (1:1 to 21:9)
- Added resolution options (1K default, 2K, 4K)
- Simplified documentation with clear defaults
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Problem:
When Claude creates worktrees, it sometimes calls `git worktree add` directly
instead of using the worktree-manager.sh script. This means .env files don't
get copied to the new worktree, causing the app to fail on startup.
Solution:
1. Add copy_env_files() function to worktree-manager.sh that copies all .env*
files (except .env.example) from main repo to new worktree
2. Call copy_env_files() automatically during worktree creation
3. Add new 'copy-env' command to manually copy env files to existing worktrees
4. Update SKILL.md with CRITICAL section instructing Claude to NEVER call
git worktree add directly - always use the manager script
5. Update all code examples to use ${CLAUDE_PLUGIN_ROOT} for portability
6. Add troubleshooting section for missing .env files
Features:
- Automatically copies .env, .env.local, .env.test, etc.
- Skips .env.example (should be in git)
- Creates .backup if destination already exists
- New 'copy-env' command for manual copying to existing worktrees
Reduced from ~645 lines to ~100 lines while keeping the core concepts:
- The seed (4 commands + universal agents)
- The growth loop via /compound
- Agent storage hierarchy
- Implementation phases
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add "Create Issue" option to post-generation menu in /plan
- Detect project tracker from user's CLAUDE.md (project_tracker: github or linear)
- Use gh issue create for GitHub, linear CLI for Linear
- Show issue URL after creation and offer to continue to /work or /plan_review
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Add CHANGELOG entry for new frontend race condition reviewer
- Update agent count from 23 to 24 in plugin.json and marketplace.json
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Added "The current year is 2025" note to all 4 research agents so they
don't assume outdated dates when searching for documentation:
- best-practices-researcher
- framework-docs-researcher
- git-history-analyzer
- repo-research-analyst
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Replace vague "keep asking questions until user is happy" ending with
AskUserQuestion tool presenting structured options. Users now see 4
explicit choices after plan generation: Start /work, Run /plan_review,
Simplify, or Rework.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Update all documentation to show the correct command syntax:
- `/plan`, `/review`, `/work`, `/codify` (short form, autocompletes)
Previously showed `/workflows:plan` etc. which doesn't work.
The `name:` field in frontmatter determines the short command name.
Fixes#19🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Applied Hunt/Thomas and Joel Spolsky writing principles:
- Concrete before abstract (stories and examples first)
- Physical analogies for technical concepts
- Conversational voice with "you" and contractions
- Removed passive voice and weasel words
- Added memorable metaphors and hooks
Pages updated:
- index.html: New hero, philosophy section with N+1 query story
- agents.html: Each agent described with concrete scenarios
- commands.html: Action-oriented descriptions
- getting-started.html: Direct, conversational guide
- skills.html: Clear "when to use this" for each skill
- mcp-servers.html: Concrete examples of what each tool does
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
GitHub Pages is configured to serve from /docs in main branch.
Moved documentation site from plugins/compounding-engineering/docs/
to repository root /docs folder.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Removed over-styled elements that were added during design iterations:
- Remove pulse, icon-float, badge-pulse, server-pulse animations
- Remove diamond-spin and slide-in animations
- Remove scanline repeating-linear-gradient effects
- Simplify hover transforms from -4px/-6px to -2px
- Simplify complex multi-layer box shadows to single shadows
- Remove shine effect on primary buttons
- Clean up duplicate/corrupted CSS rules
Net result: -205 lines of CSS, cleaner and more minimal design
following LaunchKit's "purposeful restraint" philosophy.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Adds automated deployment of documentation to GitHub Pages when
changes are pushed to the docs directory on the main branch.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
