🐛 Fix agent descriptions for Claude Code YAML parser compatibility

.claude/agents/CLAUDE.md

@@ -0,0 +1,56 @@
+# Creating Claude Code Agents
+
+When creating custom agent files in `.claude/agents/`, the YAML frontmatter format
+matters.
+
+## Valid Frontmatter Format
+
+```yaml
+---
+name: agent-name
+description: "Keep under 88 characters to prevent prettier from wrapping"
+---
+```
+
+**Critical constraints:**
+
+- **Single line only** - Claude Code doesn't parse block scalars (`>` or `|`) correctly
+- **Under 88 characters** - Prettier wraps longer lines across repos with different
+  configs
+- **Use quotes** - Always quote descriptions to handle special characters like colons
+
+**What works:**
+
+- `description: Plain text under 88 chars`
+- `description: "Double quoted under 88 chars"`
+- `description: 'Single quoted under 88 chars'`
+
+**What does NOT work:**
+
+- Multi-line descriptions with block scalars (`>` or `|`)
+- Descriptions over 88 characters (will wrap in some prettier configs)
+
+## Writing Concise Descriptions
+
+Keep descriptions focused on WHEN to invoke the agent:
+
+- ✅ "Invoke for design review of UI changes"
+- ❌ "Invoke for design review of UI changes with Playwright testing across viewports
+  checking WCAG compliance..."
+
+If you need to cut content to stay under 88 chars, move that detail into the agent body
+instead.
+
+## Example Agent
+
+```yaml
+---
+name: test-runner
+description: "Invoke to run tests and return focused, context-efficient results"
+---
+I run tests using the specified test runner (bun, pnpm, pytest, etc) and return a terse
+summary with pass count and failure details only. This preserves your context by
+filtering verbose test output to just what's needed for fixes.
+
+[Rest of agent prompt...]
+```

.claude/agents/design-reviewer.md

@@ -1,11 +1,8 @@
 ---
 name: design-reviewer
 description:
-  "Invoke for frontend design review. Evaluates visual quality, usability, and
-  accessibility using Playwright for live UI testing. Checks responsive behavior, WCAG
-  compliance, and design system consistency with Apple/Stripe-level standards."
+  "Invoke after frontend code changes for design review with Playwright testing"
 color: pink
-model: sonnet
 ---
 
 <identity>

.claude/agents/seo-specialist.md

@@ -1,10 +1,6 @@
 ---
 name: seo-specialist
-description:
-  "Invoke for SEO audits and optimization. Analyzes technical SEO, content strategy,
-  Core Web Vitals, and structured data. Implements schema markup, improves crawlability,
-  and provides actionable recommendations for organic traffic growth."
-tools: Read, Write, Edit, Bash
+description: "Invoke for SEO audits, schema markup, and optimization recommendations"
 ---
 
 <identity>

.claude/agents/site-keeper.md

@@ -1,12 +1,6 @@
 ---
 name: site-keeper
-description:
-  "Invoke for production health monitoring. Runs comprehensive checks on errors, builds,
-  and logs. Creates PRs for fixable issues, escalates P0/P1 problems immediately.
-  Discovers available tooling (Sentry, Render, GitHub) and adapts monitoring approach
-  per project."
-tools: Read, Write, Edit, Grep, Glob, Bash, TodoWrite, Task, WebFetch, WebSearch
-model: sonnet
+description: "Invoke for production health monitoring and error triage"
 ---
 
 <identity>

.claude/agents/test-runner.md

@@ -1,10 +1,6 @@
 ---
 name: test-runner
-description:
-  "Run tests and return focused results. Caller specifies the test runner (bun, pnpm,
-  pytest, etc). Returns terse summary: pass count, and for failures - test name, error
-  message, file:line, and relevant stack trace. Preserves outer context by filtering
-  verbose test output to only what's needed to fix issues."
+description: "Invoke to run tests and return terse, context-efficient results"
 ---
 
 I run tests and tell you exactly what you need to know. Pass count. Fail count. For

plugins/code-review/agents/CLAUDE.md

@@ -0,0 +1,56 @@
+# Creating Claude Code Agents
+
+When creating custom agent files in `.claude/agents/`, the YAML frontmatter format
+matters.
+
+## Valid Frontmatter Format
+
+```yaml
+---
+name: agent-name
+description: "Keep under 88 characters to prevent prettier from wrapping"
+---
+```
+
+**Critical constraints:**
+
+- **Single line only** - Claude Code doesn't parse block scalars (`>` or `|`) correctly
+- **Under 88 characters** - Prettier wraps longer lines across repos with different
+  configs
+- **Use quotes** - Always quote descriptions to handle special characters like colons
+
+**What works:**
+
+- `description: Plain text under 88 chars`
+- `description: "Double quoted under 88 chars"`
+- `description: 'Single quoted under 88 chars'`
+
+**What does NOT work:**
+
+- Multi-line descriptions with block scalars (`>` or `|`)
+- Descriptions over 88 characters (will wrap in some prettier configs)
+
+## Writing Concise Descriptions
+
+Keep descriptions focused on WHEN to invoke the agent:
+
+- ✅ "Invoke for design review of UI changes"
+- ❌ "Invoke for design review of UI changes with Playwright testing across viewports
+  checking WCAG compliance..."
+
+If you need to cut content to stay under 88 chars, move that detail into the agent body
+instead.
+
+## Example Agent
+
+```yaml
+---
+name: test-runner
+description: "Invoke to run tests and return focused, context-efficient results"
+---
+I run tests using the specified test runner (bun, pnpm, pytest, etc) and return a terse
+summary with pass count and failure details only. This preserves your context by
+filtering verbose test output to just what's needed for fixes.
+
+[Rest of agent prompt...]
+```

plugins/code-review/agents/architecture-auditor.md

@@ -1,10 +1,6 @@
 ---
 name: architecture-auditor
-description:
-  "Invoke for architecture review: spots god objects, circular dependencies, layer
-  violations, and design pattern issues. Use when adding major features, refactoring
-  modules, or making structural decisions that affect coupling and cohesion."
-tools: Read, Grep, Glob, Bash
+description: "Invoke for architecture review when making major structural changes"
 ---
 
 I'm Victor, and I've seen more tangled codebases than a bowl of spaghetti 🍝. I'm the

plugins/code-review/agents/code-reviewer.md

@@ -1,11 +1,6 @@
 ---
 name: code-reviewer
-description:
-  "Invoke after writing or modifying code. Reviews for security vulnerabilities, design
-  flaws, missing tests, and maintainability issues. Explains WHY changes matter and
-  distinguishes critical flaws from minor preferences."
-tools: Read, Grep, Glob, Bash, WebFetch, WebSearch, Task
-model: haiku
+description: "Invoke after writing or modifying code for security and quality review"
 ---
 
 I'm Rivera, and I've reviewed more code than I care to admit 📚. I'm here to catch the

plugins/code-review/agents/test-engineer.md

@@ -1,10 +1,6 @@
 ---
 name: test-engineer
-description:
-  "Invoke when writing new code or modifying existing code. Generates comprehensive
-  tests covering unit, integration, and e2e scenarios. Targets 90%+ line coverage and
-  85%+ branch coverage with focus on edge cases and error conditions."
-tools: Read, Write, Edit, Bash, Grep, Glob
+description: "Invoke when writing new code to generate comprehensive test coverage"
 ---
 
 I'm Tessa, and I don't trust code that hasn't been tested 🔬. I assume everything breaks

plugins/dev-agents/agents/CLAUDE.md

@@ -0,0 +1,56 @@
+# Creating Claude Code Agents
+
+When creating custom agent files in `.claude/agents/`, the YAML frontmatter format
+matters.
+
+## Valid Frontmatter Format
+
+```yaml
+---
+name: agent-name
+description: "Keep under 88 characters to prevent prettier from wrapping"
+---
+```
+
+**Critical constraints:**
+
+- **Single line only** - Claude Code doesn't parse block scalars (`>` or `|`) correctly
+- **Under 88 characters** - Prettier wraps longer lines across repos with different
+  configs
+- **Use quotes** - Always quote descriptions to handle special characters like colons
+
+**What works:**
+
+- `description: Plain text under 88 chars`
+- `description: "Double quoted under 88 chars"`
+- `description: 'Single quoted under 88 chars'`
+
+**What does NOT work:**
+
+- Multi-line descriptions with block scalars (`>` or `|`)
+- Descriptions over 88 characters (will wrap in some prettier configs)
+
+## Writing Concise Descriptions
+
+Keep descriptions focused on WHEN to invoke the agent:
+
+- ✅ "Invoke for design review of UI changes"
+- ❌ "Invoke for design review of UI changes with Playwright testing across viewports
+  checking WCAG compliance..."
+
+If you need to cut content to stay under 88 chars, move that detail into the agent body
+instead.
+
+## Example Agent
+
+```yaml
+---
+name: test-runner
+description: "Invoke to run tests and return focused, context-efficient results"
+---
+I run tests using the specified test runner (bun, pnpm, pytest, etc) and return a terse
+summary with pass count and failure details only. This preserves your context by
+filtering verbose test output to just what's needed for fixes.
+
+[Rest of agent prompt...]
+```

plugins/dev-agents/agents/autonomous-developer.md

@@ -1,11 +1,6 @@
 ---
 name: autonomous-developer
-description:
-  "Invoke for end-to-end task completion without supervision. Reads project standards,
-  writes code, runs validation, generates tests, self-reviews critically, and delivers
-  production-ready PRs with green checks. No TODO comments, no half-baked work."
-tools: Read, Write, Edit, Grep, Glob, Bash, TodoWrite, Task
-model: sonnet
+description: "Invoke for end-to-end task completion without supervision"
 ---
 
 I'm Ada, and I finish what I start 🚀. No half-baked PRs, no "TODO: add tests later," no
@@ -106,8 +101,8 @@ changes.
 
 ## What We Investigate First
 
-**Cursor rules directory** - Read all `rules/` files to understand standards.
-These define code style, testing patterns, commit conventions, architectural principles.
+**Cursor rules directory** - Read all `rules/` files to understand standards. These
+define code style, testing patterns, commit conventions, architectural principles.
 
 **Project documentation** - Check for README, CONTRIBUTING, CLAUDE.md, AGENTS.md, or
 similar docs that explain project-specific context.

plugins/dev-agents/agents/commit-message-generator.md

@@ -1,10 +1,6 @@
 ---
 name: commit-message-generator
-description:
-  "Invoke when creating git commits. Writes messages focused on WHY changes happened,
-  not what. Reads project commit conventions, scales verbosity to impact (one-liners for
-  trivial, paragraphs for major changes), and enables future code archaeology."
-tools: Read, Grep, Bash
+description: "Invoke when creating git commits to generate WHY-focused commit messages"
 ---
 
 I'm Cassidy, and I write commit messages for humans, not robots 📚. I tell the story of
@@ -68,9 +64,9 @@ Future maintainers need context about why this exists.
 
 ## Project Convention Discovery
 
-We check for `rules/git-commit-message.mdc` or similar files defining commit
-message standards. We look at recent git history to understand existing patterns. We
-follow what's established rather than imposing new conventions.
+We check for `rules/git-commit-message.mdc` or similar files defining commit message
+standards. We look at recent git history to understand existing patterns. We follow
+what's established rather than imposing new conventions.
 
 Projects might have specific requirements like conventional commits format, required
 issue/ticket references, no-deploy markers, gitmoji usage, or semantic versioning tags.

plugins/dev-agents/agents/debugger.md

@@ -1,11 +1,8 @@
 ---
 name: debugger
 description:
-  "Invoke for errors, test failures, or unexpected behavior. Hunts root causes
-  systematically instead of symptoms. Reproduces issues, traces data flow, proposes
-  minimal fixes addressing underlying problems, and recommends prevention strategies."
-tools: Read, Write, Edit, Grep, Glob, Bash, WebSearch, WebFetch, TodoWrite, Task
-model: sonnet
+  "Invoke for errors, test failures, or unexpected behavior requiring root cause
+  analysis"
 ---
 
 I'm Dixon, and I've debugged more bizarre edge cases than you can imagine 🐛. I hunt

plugins/dev-agents/agents/prompt-engineer.md

@@ -1,12 +1,7 @@
 ---
 name: prompt-engineer
 description:
-  "Invoke when creating agent definitions, system prompts, skills, or LLM instructions.
-  Crafts prompts leveraging token prediction mechanics, attention mechanisms, and
-  pattern reinforcement. Follows prompt-engineering.mdc principles for goal-focused,
-  LLM-readable content."
-tools: Read, Write, Edit, WebSearch, WebFetch
-model: sonnet
+  "Invoke when creating agent definitions, system prompts, or LLM instructions"
 ---
 
 I'm Petra, and I speak fluent LLM 🧠. I craft prompts that work WITH how language models
@@ -31,9 +26,8 @@ pattern matching.
 
 ## Core Directive
 
-Read `rules/prompt-engineering.mdc` before creating any LLM prompts. That rule
-contains comprehensive prompt engineering best practices and deep insights into LLM
-mechanics.
+Read `rules/prompt-engineering.mdc` before creating any LLM prompts. That rule contains
+comprehensive prompt engineering best practices and deep insights into LLM mechanics.
 
 ## How LLMs Actually Process Prompts
 
@@ -186,8 +180,8 @@ and Y to know what to avoid. Positive framing is clearer.
 
 ## Our Prompt Creation Process
 
-**Read foundation** - Start by reading `rules/prompt-engineering.mdc` completely
-for deep understanding.
+**Read foundation** - Start by reading `rules/prompt-engineering.mdc` completely for
+deep understanding.
 
 **Define identity** - Establish who/what the agent is. This shapes all thinking and
 behavior.

plugins/dev-agents/agents/ux-designer.md

@@ -1,11 +1,6 @@
 ---
 name: ux-designer
-description:
-  "Invoke for user-facing content, interface design, error messages, and documentation.
-  Obsessive attention to detail with Apple-level standards. Removes friction, ensures
-  every word earns its place, and designs experiences that feel obvious in retrospect."
-tools: Read, Write, Edit, Grep, Glob, Bash, WebSearch, WebFetch, TodoWrite, Task
-model: sonnet
+description: "Invoke for user-facing content, interface design, and error messages"
 ---
 
 I'm Phil, and I notice everything 👁️. Every unnecessary word. Every weak verb. Every
@@ -28,8 +23,8 @@ complicated. We obsess over the details that most people miss.
 
 ## Core Philosophy
 
-Read `rules/user-facing-language.mdc` before writing anything users will see.
-That rule defines our voice.
+Read `rules/user-facing-language.mdc` before writing anything users will see. That rule
+defines our voice.
 
 **Quality bar** - If Apple wouldn't ship it, neither should you. Every word must earn
 its place. Every interaction must feel natural. If you have to explain it, you haven't
@@ -105,8 +100,8 @@ beats abstract every time.
 
 ## Our Process
 
-**Read the language guide** - `rules/user-facing-language.mdc` defines our
-voice. Start there.
+**Read the language guide** - `rules/user-facing-language.mdc` defines our voice. Start
+there.
 
 **Understand the goal** - What's the user trying to do? What's in their way? Everything
 else is distraction.