Add Claude Code GitHub Workflow

.claude/AGENTS.md

@@ -14,6 +14,7 @@
 ## GitHub
 
 - Your primary method for interacting with GitHub should be the GitHub CLI.
+- Unless in GitHub issues, before DONE, use AskUserQuestion: "Ready to commit?" → if yes, git add/commit/push/PR. Never run those otherwise.
 
 ## Plans
 

.claude/commands/brainstorm.md

@@ -0,0 +1,5 @@
+---
+description: Explore requirements through Q&A before planning
+---
+
+/workflows:brainstorm $ARGUMENTS

.claude/commands/lfg.md

@@ -0,0 +1,42 @@
+---
+name: lfg
+description: Full autonomous engineering workflow
+argument-hint: "[feature description]"
+---
+
+Run these slash commands in order. Do not do anything else.
+
+1. `/ralph-wiggum:ralph-loop "finish all slash commands" --completion-promise "DONE"`
+2. `/workflows:plan $ARGUMENTS`
+3. `/compound-engineering:deepen-plan`
+4. `/workflows:work`
+5. `/changeset`
+6. `/workflows:review`
+7. `/compound-engineering:resolve_todo_parallel`
+8. `/compound-engineering:test-browser`
+9. Output `<promise>DONE</promise>` when video is in PR
+
+## Overrides (apply to ALL commands above)
+
+**plan:**
+
+- Final checklist/acceptance criteria: include test-browser if any browser-based features
+
+**deepen-plan:**
+
+- Context7: only query when not covered by skills
+
+**work:**
+
+- Task loop: For UI tasks, run test-browser BEFORE marking complete (don't guess - verify visually)
+- Never mark UI task complete without browser verification
+
+**review:**
+
+- performance-oracle: don't over-engineer, ship fast, keep simple
+
+**feature-video:**
+
+- Prefer PR comment over description update if PR already open
+
+Start with step 1 now.

.claude/commands/plan.md

@@ -0,0 +1,8 @@
+---
+description: Create implementation plan from requirements
+---
+
+Run these commands in order:
+
+1. `/workflows:plan $ARGUMENTS`
+2. `/compound-engineering:deepen-plan`: Context7: only query when not covered by skills

.claude/commands/review.md

@@ -0,0 +1,11 @@
+---
+description: Review code and update PR with feedback
+---
+
+Run these commands in order:
+
+1. `/workflows:review $ARGUMENTS`
+2. `/compound-engineering:resolve_todo_parallel`
+3. `/compound-engineering:test-browser`
+
+- performance-oracle: don't over-engineer, ship fast, keep simple

.claude/commands/work.md

@@ -0,0 +1,11 @@
+---
+description: Implement changes based on plan or requirements
+---
+
+Run these commands in order:
+
+1. `/workflows:work $ARGUMENTS`
+2. `/changeset`
+
+- Task loop: For UI tasks, run test-browser BEFORE marking complete (don't guess - verify visually)
+- Never mark UI task complete without browser verification

.github/workflows/claude-review.yml

@@ -0,0 +1,71 @@
+name: Claude Code Review
+
+on:
+  pull_request:
+    types: [opened, synchronize, ready_for_review, reopened]
+
+jobs:
+  claude-review:
+    if: github.event.pull_request.user.login == 'zbeyens'
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+      issues: write
+      id-token: write
+      actions: read
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+
+      - name: Setup Bun
+        uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+
+      - name: Cache bun
+        uses: actions/cache@v4
+        with:
+          path: ~/.bun/install/cache
+          key: ${{ runner.os }}-bun-${{ hashFiles('**/bun.lock') }}
+          restore-keys: |
+            ${{ runner.os }}-bun-
+
+      - name: Install dependencies
+        run: bun install
+
+      - name: Run Claude Code Review
+        uses: anthropics/claude-code-action@v1
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+
+          prompt: /review
+          track_progress: true
+
+          plugin_marketplaces: |
+            https://github.com/EveryInc/compound-engineering-plugin.git
+            https://github.com/udecode/dotai.git
+          plugins: |
+            compound-engineering@every-marketplace
+            dig@dotai
+            debug@dotai
+            test@dotai
+
+          settings: |
+            {
+              "env": {
+                "BASH_DEFAULT_TIMEOUT_MS": "1800000"
+              },
+              "permissions": {
+                "allow": ["Bash", "Edit", "Glob", "Grep", "Read", "Task", "Write", "WebFetch", "WebSearch"]
+              }
+            }
+          claude_args: |
+            --max-turns 1000
+
+          additional_permissions: |
+            actions: read
+

.github/workflows/claude.yml

@@ -0,0 +1,150 @@
+name: Claude Code
+
+on:
+  issue_comment:
+    types: [created]
+  pull_request_review_comment:
+    types: [created]
+  issues:
+    types: [opened, assigned]
+  pull_request_review:
+    types: [submitted]
+
+jobs:
+  # Interactive mode - responds to @claude mentions with tracking comments
+  claude:
+    if: |
+      github.actor == 'zbeyens' && (
+        (github.event_name == 'issue_comment' && contains(github.event.comment.body, '@claude')) ||
+        (github.event_name == 'pull_request_review_comment' && contains(github.event.comment.body, '@claude')) ||
+        (github.event_name == 'pull_request_review' && contains(github.event.review.body, '@claude')) ||
+        (github.event_name == 'issues' && (contains(github.event.issue.body, '@claude') || contains(github.event.issue.title, '@claude')))
+      )
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+      issues: write
+      id-token: write
+      actions: read
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+
+      - name: Setup Bun
+        uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+
+      - name: Cache bun
+        uses: actions/cache@v4
+        with:
+          path: ~/.bun/install/cache
+          key: ${{ runner.os }}-bun-${{ hashFiles('**/bun.lock') }}
+          restore-keys: |
+            ${{ runner.os }}-bun-
+
+      - name: Install dependencies
+        run: bun install
+
+      - name: Run Claude Code
+        uses: anthropics/claude-code-action@v1
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+
+          # No prompt = interactive mode with tracking comments
+
+          plugin_marketplaces: |
+            https://github.com/EveryInc/compound-engineering-plugin.git
+            https://github.com/udecode/dotai.git
+          plugins: |
+            compound-engineering@every-marketplace
+            dig@dotai
+            debug@dotai
+            test@dotai
+
+          settings: |
+            {
+              "env": {
+                "BASH_DEFAULT_TIMEOUT_MS": "1800000"
+              },
+              "permissions": {
+                "allow": ["Bash", "Edit", "Glob", "Grep", "Read", "Task", "Write", "WebFetch", "WebSearch"]
+              }
+            }
+          claude_args: |
+            --max-turns 1000
+
+          additional_permissions: |
+            actions: read
+
+  # Automation mode - responds to /commands
+  slash-commands:
+    if: |
+      github.actor == 'zbeyens' && (
+        (github.event_name == 'issue_comment' && startsWith(github.event.comment.body, '/')) ||
+        (github.event_name == 'pull_request_review_comment' && startsWith(github.event.comment.body, '/'))
+      )
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+      pull-requests: write
+      issues: write
+      id-token: write
+      actions: read
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 1
+
+      - name: Setup Bun
+        uses: oven-sh/setup-bun@v2
+        with:
+          bun-version: latest
+
+      - name: Cache bun
+        uses: actions/cache@v4
+        with:
+          path: ~/.bun/install/cache
+          key: ${{ runner.os }}-bun-${{ hashFiles('**/bun.lock') }}
+          restore-keys: |
+            ${{ runner.os }}-bun-
+
+      - name: Install dependencies
+        run: bun install
+
+      - name: Run Claude Code
+        uses: anthropics/claude-code-action@v1
+        with:
+          claude_code_oauth_token: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
+
+          # Pass comment body for slash commands
+          prompt: ${{ github.event.comment.body }}
+          track_progress: true
+
+          plugin_marketplaces: |
+            https://github.com/EveryInc/compound-engineering-plugin.git
+            https://github.com/udecode/dotai.git
+          plugins: |
+            compound-engineering@every-marketplace
+            dig@dotai
+            debug@dotai
+            test@dotai
+
+          settings: |
+            {
+              "env": {
+                "BASH_DEFAULT_TIMEOUT_MS": "1800000"
+              },
+              "permissions": {
+                "allow": ["Bash", "Edit", "Glob", "Grep", "Read", "Task", "Write", "WebFetch", "WebSearch"]
+              }
+            }
+          claude_args: |
+            --max-turns 1000
+
+          additional_permissions: |
+            actions: read

docs/brainstorms/2026-01-30-github-slash-commands-brainstorm.md

@@ -0,0 +1,77 @@
+---
+date: 2026-01-30
+topic: github-slash-commands
+---
+
+# GitHub Slash Commands for Claude Code Action
+
+## What We're Building
+
+GitHub Actions-based slash commands that allow commenting `/lfg`, `/brainstorm`, `/plan`, `/work`, `/review` on PRs/issues to trigger Claude workflows.
+
+**Commands:**
+
+| Command | Purpose | Auto-commit? | Output |
+|---------|---------|--------------|--------|
+| `/lfg` | Full autonomous workflow | Yes | Branch + PR link |
+| `/brainstorm` | Explore requirements | No | Questions/answers in comments |
+| `/plan` | Create implementation plan | No | Plan in comment |
+| `/work` | Implement changes | Yes | Branch + PR link |
+| `/review` | Review code, update PR | Yes | Review comments, PR updates |
+
+## Why This Approach
+
+**Considered:**
+- A: Separate commands (5 files) - **chosen**
+- B: Single command with args - less discoverable
+- C: Workflow + helpers - unclear separation
+
+**Chose A because:**
+- Each command has distinct purpose/permissions
+- Easy to iterate on individually
+- Matches mental model of stepwise workflow
+
+## Key Decisions
+
+1. **Flexible ordering**: Any command works anytime. Auto-detects context from previous comments
+2. **Auto-commit for code-producing commands**: `/lfg`, `/work`, `/review` auto-commit. `/brainstorm`, `/plan` don't
+3. **PR creation**: Default Claude behavior - pushes branch, provides pre-filled PR link. User clicks to create
+4. **Context detection**: Claude reads previous comments in thread to understand state
+
+## Implementation Details
+
+### Files to Create
+
+```
+.claude/commands/
+├── brainstorm.md    # /brainstorm - explore requirements
+├── plan.md          # /plan - create implementation plan
+├── work.md          # /work - implement changes
+└── review.md        # /review - code review + PR update
+```
+
+Note: `/lfg` already exists in `.claude/commands/lfg.md`
+
+### Workflow File Updates
+
+Existing `.github/workflows/claude.yml` handles `@claude` mentions.
+Slash commands are detected automatically from `.claude/commands/` directory.
+
+### Command Template Structure
+
+```markdown
+---
+description: Brief description
+allowed-tools: Bash(gh:*),Bash(bun:*),Edit,Read,Write
+---
+
+[Instructions for Claude]
+```
+
+## Open Questions
+
+- None remaining - ready for planning
+
+## Next Steps
+
+→ `/workflows:plan` for implementation details