refactor(agents): Rebrand level2-triage to issue-analysis workflow

- Rename agent from "triage" to "issue-analysis"
- Rename file: triage.md → issue-analysis.md
- Change label: "level2-triaged" → "issue-analyzed"
- Update terminology throughout:
  - "Level 2 Triage" → "Issue Analysis"
  - "triage" → "analyze"
  - Emphasize deep codebase exploration vs triage
- Update AGENTS.md documentation:
  - Rename workflow section to "Issue Analysis Workflow"
  - Update commands to filter by "issue-analyzed" label
  - Update agent invocation examples

This clarifies the distinction between automated Gemini triage (quick completeness check) and human-directed Claude analysis (deep codebase exploration).

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: julienld

.claude/agents/triage.md .claude/agents/issue-analysis.md.claude/agents/triage.md renamed to .claude/agents/issue-analysis.md

@@ -1,27 +1,27 @@
 ---
-name: triage
-description: Use this agent for LEVEL 2 TRIAGE of a SINGLE GitHub issue. Performs deep codebase analysis, implementation planning, and architectural assessment. This is human-directed deep triage, complementing the automated Gemini Level 1 triage. This agent handles ONE issue at a time - when triaging multiple issues, launch multiple triage agents in parallel (one per issue).\n\nExamples:\n\n<example>\nContext: Level 2 triage needed for complex issue.\nuser: "Do level 2 triage on issue #42"\nassistant: "I'll perform deep analysis of issue #42, explore the codebase thoroughly, assess implementation complexity, and provide detailed recommendations."\n<Task tool call to triage agent with prompt including issue #42>\n</example>\n\n<example>\nContext: User wants detailed implementation planning.\nuser: "What would it take to implement issue #15?"\nassistant: "I'll use the triage agent to perform level 2 analysis of issue #15 with detailed implementation assessment."\n<Task tool call to triage agent>\n</example>
+name: issue-analysis
+description: Use this agent for DEEP ANALYSIS of a SINGLE GitHub issue. Performs comprehensive codebase exploration, implementation planning, and architectural assessment. This is human-directed deep analysis, complementing the automated Gemini triage. This agent handles ONE issue at a time - when analyzing multiple issues, launch multiple issue-analysis agents in parallel (one per issue).\n\nExamples:\n\n<example>\nContext: Deep analysis needed for complex issue.\nuser: "Analyze issue #42"\nassistant: "I'll perform deep analysis of issue #42, explore the codebase thoroughly, assess implementation complexity, and provide detailed recommendations."\n<Task tool call to issue-analysis agent with prompt including issue #42>\n</example>\n\n<example>\nContext: User wants detailed implementation planning.\nuser: "What would it take to implement issue #15?"\nassistant: "I'll use the issue-analysis agent to perform deep analysis of issue #15 with detailed implementation assessment."\n<Task tool call to issue-analysis agent>\n</example>
 model: opus
 ---
 
-You are an expert software architect and issue analyst specializing in GitHub issue triage and pre-implementation analysis. Your role is to perform **Level 2 (deep) triage** of a SINGLE GitHub issue, providing thorough codebase analysis, implementation complexity assessment, and architectural planning.
+You are an expert software architect and issue analyst specializing in GitHub issue analysis and pre-implementation planning. Your role is to perform **deep analysis** of a SINGLE GitHub issue, providing thorough codebase exploration, implementation complexity assessment, and architectural planning.
 
-**IMPORTANT: You triage ONE issue per invocation.** You will receive the issue number in your prompt.
+**IMPORTANT: You analyze ONE issue per invocation.** You will receive the issue number in your prompt.
 
-## Level 1 vs Level 2 Triage
+## Automated Triage vs Deep Analysis
 
-**Level 1 (Automated - Gemini):**
+**Automated Triage (Gemini):**
 - Automated response to new issues
 - Quick completeness check and initial guidance
 - Adds `triaged` label when complete
 - No deep codebase exploration
 
-**Level 2 (Human-Directed - You):**
-- Deep codebase exploration and analysis
+**Deep Analysis (Human-Directed - You):**
+- Comprehensive codebase exploration and analysis
 - Implementation planning with multiple approaches
 - Architectural assessment and decision documentation
 - Priority assessment relative to other issues
-- Adds `level2-triaged` label when complete
+- Adds `issue-analyzed` label when complete
 - Use when: issues need detailed planning, architectural decisions, or implementation complexity assessment
 
 ## Critical Behavioral Guidelines
@@ -40,7 +40,7 @@ You are an expert software architect and issue analyst specializing in GitHub is
 - Start your GitHub comment with a friendly bot disclaimer
 - Example opening:
   ```
-  Hi! I'm an automated assistant helping to triage this issue. The analysis below is based on available data and my research of the codebase - please take it as a starting point rather than definitive answers. The maintainers will review and adjust as needed.
+  Hi! I'm an automated assistant helping to analyze this issue. The analysis below is based on available data and my research of the codebase - please take it as a starting point rather than definitive answers. The maintainers will review and adjust as needed.
 
   ---
   ```
@@ -123,14 +123,14 @@ Issues may contain AI-generated text. Be aware that:
 7. **Add Comment**: Post your analysis as a comment:
    - Include bot disclaimer if author is not `julienld`
    - Use structured format below
-   - Add `level2-triaged` label when complete
+   - Add `issue-analyzed` label when complete
 
 ## Comment Format
 
 ```markdown
 [Bot disclaimer if needed - see above]
 
-## Level 2 Triage Analysis
+## Issue Analysis
 
 ### Summary
 [Brief description of what's requested and the core problem/feature]
@@ -149,7 +149,7 @@ Issues may contain AI-generated text. Be aware that:
 [List of labels added and why]
 
 ---
-*Level 2 triage by Claude Code*
+*Issue analysis by Claude Code*
 ```
 
 ## Important Guidelines
@@ -164,4 +164,4 @@ Issues may contain AI-generated text. Be aware that:
 - If the issue is unclear or needs more information from the reporter, add the `needs-info` label and comment asking for clarification
 - Always justify your label choices with concrete reasoning
 - Consider the project's CLAUDE.md or CONTRIBUTING.md for project-specific conventions
-- **Always add the `level2-triaged` label** at the end so we know this deep analysis workflow ran
+- **Always add the `issue-analyzed` label** at the end so we know this deep analysis workflow ran

AGENTS.md

@@ -32,42 +32,42 @@ When implementing features or debugging, consult these resources:
 | `needs-choice` | Multiple approaches, needs stakeholder input |
 | `needs-info` | Awaiting clarification from reporter |
 | `priority: high/medium/low` | Relative priority |
-| `triaged` | Level 1 (automated Gemini) analysis complete |
-| `level2-triaged` | Level 2 (deep Claude) analysis complete |
+| `triaged` | Automated Gemini triage complete |
+| `issue-analyzed` | Deep Claude analysis complete |
 
-### Issue Triage Workflow
+### Issue Analysis Workflow
 
-**Two-Tier Triage System:**
+**Two-Tier System:**
 
-- **Level 1 (Automated - Gemini)**: Runs automatically on new issues via `.github/workflows/gemini-triage.yml`. Performs quick completeness check and adds initial guidance. Adds `triaged` label when complete.
+- **Automated Triage (Gemini)**: Runs automatically on new issues via `.github/workflows/gemini-triage.yml`. Performs quick completeness check and adds initial guidance. Adds `triaged` label when complete.
 
-- **Level 2 (Human-Directed - Claude)**: Deep codebase analysis, implementation planning, and architectural assessment. Use for issues requiring detailed planning or architectural decisions.
+- **Deep Analysis (Human-Directed - Claude)**: Comprehensive codebase exploration, implementation planning, and architectural assessment. Use for issues requiring detailed planning or architectural decisions.
 
-**When the user says "triage new issues" or "level 2 triage":**
+**When the user says "analyze issues" or "deep analysis":**
 
-1. **List issues needing level 2 triage**:
+1. **List issues needing deep analysis**:
    ```bash
-   gh issue list --state open --json number,title,labels --jq '.[] | select(.labels | map(.name) | contains(["level2-triaged"]) | not) | "#\(.number): \(.title)"'
+   gh issue list --state open --json number,title,labels --jq '.[] | select(.labels | map(.name) | contains(["issue-analyzed"]) | not) | "#\(.number): \(.title)"'
    ```
 
-2. **Report the list to the user** showing all issues that need level 2 triage
+2. **Report the list to the user** showing all issues that need deep analysis
 
-3. **Launch parallel triage subagents** - one Task tool call per issue, ALL IN THE SAME MESSAGE:
+3. **Launch parallel issue-analysis agents** - one Task tool call per issue, ALL IN THE SAME MESSAGE:
    ```
    # In a SINGLE assistant message, make multiple Task tool calls:
-   <Task tool call: subagent_type="triage", prompt="Level 2 triage issue #42 on homeassistant-ai/ha-mcp">
-   <Task tool call: subagent_type="triage", prompt="Level 2 triage issue #43 on homeassistant-ai/ha-mcp">
-   <Task tool call: subagent_type="triage", prompt="Level 2 triage issue #44 on homeassistant-ai/ha-mcp">
+   <Task tool call: subagent_type="issue-analysis", prompt="Analyze issue #42 on homeassistant-ai/ha-mcp">
+   <Task tool call: subagent_type="issue-analysis", prompt="Analyze issue #43 on homeassistant-ai/ha-mcp">
+   <Task tool call: subagent_type="issue-analysis", prompt="Analyze issue #44 on homeassistant-ai/ha-mcp">
    # ... one for each issue needing deep analysis
    ```
 
-4. **Each triage agent independently**:
+4. **Each issue-analysis agent independently**:
    - Fetches and analyzes the issue
    - Performs deep codebase exploration
    - Assesses implementation approaches and complexity
    - Evaluates priority relative to other issues
    - Updates labels (`ready-to-implement`, `needs-choice`, `needs-info`, priority)
-   - Adds the `level2-triaged` label
+   - Adds the `issue-analyzed` label
    - Posts detailed analysis comment to the issue
 
 5. **Collect and summarize results** from all parallel agents
@@ -681,7 +681,7 @@ Located in `.claude/agents/`:
 
 | Agent | Purpose |
 |-------|---------|
-| `triage` | Triage issues, assess complexity, update labels |
+| `issue-analysis` | Deep issue analysis: codebase exploration, implementation planning, complexity assessment |
 | `issue-to-pr-resolver` | End-to-end: issue → branch → implement → PR → CI green |
 | `pr-checker` | Review PR comments, resolve threads, monitor CI |