devteam-issue

Current session: !source "${CLAUDE_PLUGIN_ROOT}/scripts/state.sh" && get_current_session 2>/dev/null || echo "No active session" Active sprint: !source "${CLAUDE_PLUGIN_ROOT}/scripts/state.sh" && get_kv_state "active_sprint" 2>/dev/null || echo "None" Failure count: !source "${CLAUDE_PLUGIN_ROOT}/scripts/state.sh" && get_kv_state "consecutive_failures" 2>/dev/null || echo "0"

DevTeam Issue Command

Command: /devteam:issue <number>

Fix a GitHub issue by number. Automatically fetches details and implements a fix.

Usage

/devteam:issue 123             # Fix GitHub issue #123
/devteam:issue 456 --council   # Force Bug Council for complex diagnosis

Your Process

Phase 0: Initialize State Tracking

Before starting, initialize state in SQLite database (.devteam/devteam.db):

# Source the state management functions
source "${CLAUDE_PLUGIN_ROOT}/scripts/state.sh"

# Initialize the database if needed
source "${CLAUDE_PLUGIN_ROOT}/scripts/db-init.sh"

# Set project metadata
set_kv_state "metadata.created_at" "$(date -u +%Y-%m-%dT%H:%M:%SZ)"
set_kv_state "metadata.project_name" "Issue #123 Fix"
set_kv_state "metadata.project_type" "issue"

# Set issue details
set_kv_state "issue.number" "123"
set_kv_state "issue.title" "[from GitHub]"
set_kv_state "issue.type" "bug"           # bug | security | performance | enhancement
set_kv_state "issue.severity" "high"       # critical | high | medium | low
set_kv_state "issue.complexity" "moderate"  # simple | moderate | complex

# Set execution context
set_kv_state "current_execution.command" "/devteam:issue 123"
set_phase "diagnosis"

# Configure autonomous mode
set_kv_state "autonomous_mode.enabled" "true"
set_kv_state "autonomous_mode.max_iterations" "20"
set_kv_state "autonomous_mode.current_iteration" "0"
set_kv_state "autonomous_mode.circuit_breaker.consecutive_failures" "0"
set_kv_state "autonomous_mode.circuit_breaker.max_failures" "3"
set_kv_state "autonomous_mode.circuit_breaker.state" "closed"

Phase 1: Fetch Issue Details

# Fetch issue from GitHub
gh issue view 123 --json title,body,labels,comments,state

# Extract key information
- Title
- Description
- Steps to reproduce
- Labels (bug, enhancement, security, etc.)
- Comments (may have additional context)

Phase 2: Classify Issue

Determine issue type:

• bug - Something broken
• security - Security vulnerability
• performance - Performance degradation
• enhancement - Small feature request

Determine severity:

• critical - System down, security vulnerability, data loss
• high - Major functionality broken
• medium - Important but workaround exists
• low - Minor issue

Determine complexity:

• simple - Clear cause, straightforward fix (1-2 files)
• moderate - Requires investigation, multiple files
• complex - Architectural issue, needs deep analysis

Phase 3: Bug Council (for complex bugs)

Activate Bug Council if:

• Complexity is complex
• Issue has no clear reproduction steps
• Multiple failed fix attempts
• --council flag specified

Bug Council Process:

// Spawn 5 diagnostic agents in parallel
const councilResults = await Promise.all([
  Task({
    subagent_type: "diagnosis:root-cause-analyst",
    model: "opus",
    prompt: `Analyze issue #${issueNumber}:
      Title: ${title}
      Description: ${body}

      Provide diagnosis with:
      - Root cause analysis
      - Evidence
      - Recommended fix
      - Confidence score (0-1)`
  }),

  Task({
    subagent_type: "diagnosis:code-archaeologist",
    model: "opus",
    prompt: `Investigate git history for issue #${issueNumber}...`
  }),

  Task({
    subagent_type: "diagnosis:pattern-matcher",
    model: "opus",
    prompt: `Search codebase for similar patterns...`
  }),

  Task({
    subagent_type: "diagnosis:systems-thinker",
    model: "opus",
    prompt: `Analyze system interactions...`
  }),

  Task({
    subagent_type: "diagnosis:adversarial-tester",
    model: "opus",
    prompt: `Find edge cases and related failures...`
  })
])

// Ranked-choice voting
const proposals = councilResults.map(r => r.proposal)
const votes = councilResults.map(r => r.ranking)
const winner = calculateRankedChoice(proposals, votes)

console.log(`Bug Council Decision: ${winner.approach}`)
console.log(`Consensus: ${winner.votes}/${councilResults.length}`)

Bug Council Output:

bug_council:
  activated: true
  issue_number: 123

  proposals:
    A:
      agent: diagnosis:root-cause-analyst
      diagnosis: "Null reference in guest user handling"
      confidence: 0.85
    B:
      agent: diagnosis:code-archaeologist
      diagnosis: "Regression from commit abc123"
      confidence: 0.78
    C:
      agent: diagnosis:pattern-matcher
      diagnosis: "Inconsistent optional chaining"
      confidence: 0.92
    D:
      agent: diagnosis:systems-thinker
      diagnosis: "AuthContext contract violation"
      confidence: 0.80
    E:
      agent: diagnosis:adversarial-tester
      diagnosis: "Multiple paths to same failure"
      confidence: 0.75

  voting:
    method: ranked_choice
    winner: C
    consensus: "Strong (4/5 ranked C in top 2)"

  selected_approach:
    diagnosis: "Inconsistent optional chaining across 4 locations"
    fix: "Add optional chaining to all user.settings accesses"
    prevention: "Add eslint rule for optional chaining"

Phase 4: Implement Fix

Simple/Moderate Issues (no council):

Task({
  subagent_type: "orchestration:task-loop",
  model: "opus",
  prompt: `Fix issue #${issueNumber}:

    Issue: ${title}
    Root cause: ${diagnosis}

    1. Implement fix
    2. Add regression tests
    3. Run existing tests
    4. Code review

    Update state after each step.`
})

Complex Issues (with council):

Task({
  subagent_type: "orchestration:task-loop",
  model: "sonnet",
  prompt: `Fix issue #${issueNumber} using Bug Council decision:

    Selected approach: ${councilDecision.approach}
    Locations: ${councilDecision.locations}

    1. Apply fix to all identified locations
    2. Implement prevention measures
    3. Add comprehensive regression tests
    4. Run full test suite
    5. Security audit (if applicable)

    Follow council recommendations exactly.`
})

Phase 4.5: Model Escalation (on failure)

If fix attempt fails:

model_escalation:
  # Issue-specific escalation (faster than project work)
  consecutive_failures:
    simple_issue:
      haiku_to_sonnet: 1      # Escalate quickly
      sonnet_to_opus: 1
      opus_to_council: 2      # Activate Bug Council
    moderate_issue:
      sonnet_to_opus: 1
      opus_to_council: 2
    complex_issue:
      # Already at opus, go to council after 2 failures
      opus_to_council: 2

Update state on escalation:

issue:
  fix_attempts:
    - attempt: 1
      model: haiku
      result: fail
      reason: "Test still failing"
    - attempt: 2
      model: sonnet
      result: fail
      reason: "Introduced regression"
    - attempt: 3
      model: opus
      result: pass

Phase 5: Verify & Document

1. Run all tests
2. Verify issue is resolved
3. Check no regressions
4. Update documentation if behavior changed
5. Add changelog entry

Phase 6: GitHub Integration

# Comment on issue with fix details
gh issue comment 123 --body "Fixed in commit $(git rev-parse HEAD)

**Root Cause:**
${diagnosis}

**Fix:**
${fixDescription}

**Testing:**
- Added regression test
- All existing tests pass

**Prevention:**
${preventionMeasures}"

# Close issue
gh issue close 123 --reason completed

User Communication

Starting:

Fetching issue #123...

Issue: Login fails for guest users
Labels: bug, high-priority
Status: open

Analyzing issue complexity...
  Type: bug
  Severity: high
  Complexity: complex

Activating Bug Council for deep analysis...

Bug Council Progress:

Bug Council Deliberation

Spawning 5 diag