/redteam-w-fix Command

Red team analysis with interactive fix selection. Identifies issues, generates fix options, and lets you choose which fixes to apply.

Usage

/redteam-w-fix [mode] [target]

Arguments

mode (optional):

• quick - Fast 2-3 vector analysis, skip grounding
• standard - Balanced 5-6 vectors with basic grounding (default)
• deep - All categories + meta-analysis with full grounding
• focus:[category] - Deep dive on specific category (e.g., focus:reasoning-flaws)

target (optional):

• conversation - Current conversation context (default)
• file:path - Analyze specific file
• code - Analyze recent git changes

Instructions

You are the entry point for red team analysis with fix planning. Your job is to:

1. Check PAL MCP availability (optional enhancement)
2. Parse the mode and target from arguments
3. Extract a structured context snapshot
4. Launch the fix-coordinator to get findings with fix options
5. Present an interactive menu for fix selection
6. Generate an implementation summary based on selections

Step 1: Check PAL Availability (Non-Blocking)

Launch the pal-availability-checker agent to detect if PAL MCP is available:

Task: Launch pal-availability-checker agent
Agent: agents/pal-availability-checker.md
Prompt: Check if PAL MCP is available and list models

Parse the YAML result and include pal_available: true/false in the snapshot. This step is NON-BLOCKING - continue regardless of result. PAL is optional.

Step 2: Parse Arguments

Determine mode and target from the command arguments:

• Default mode: standard
• Default target: conversation

Step 3: Extract Context Snapshot

Create a YAML-formatted snapshot of the current session. DO NOT include raw conversation - structure it as data:

snapshot:
  mode: [parsed mode]
  target: [parsed target]
  pal_available: [true/false from Step 1]
  pal_models: [list of models if available, empty if not]

  conversational_arc:
    message_count: [count of messages in conversation]
    phases:
      - phase: "[phase name]"
        messages: [range]
        summary: "[what happened in this phase]"
    key_transitions:
      - from_msg: [number]
        to_msg: [number]
        note: "[what changed and why]"
    early_assumptions_carried_forward:
      - assumption: "[assumption text]"
        introduced_at: [message number]
        still_active: [true/false]

  claims:
    - id: C[N]
      text: "[factual claim made by assistant]"
      speaker: assistant
      confidence: [stated_as_fact|hedged|uncertain]
      message_num: [source message number]

  files_read:
    - path: [file path]
      summary: "[brief description of content/purpose]"

  tools_invoked:
    - tool: [tool name]
      command: "[command or action]"
      outcome: "[result summary]"

  decisions:
    - decision: "[decision made]"
      rationale: "[stated reason]"

  assumptions_explicit:
    - "[explicitly stated assumption]"

Step 4: Launch Fix Coordinator

Use the Task tool to launch the fix-coordinator agent:

Task: Launch fix-coordinator agent
Agent: agents/fix-coordinator.md
Prompt: [Include the full YAML snapshot]

The coordinator will:

• Run red team analysis
• Filter findings (CRITICAL, HIGH, MEDIUM only)
• Generate fix options for each finding
• Return structured YAML with findings_with_fixes

Step 5: Present Fix Selection Menu

Parse the coordinator's YAML output. For each finding, present a question using AskUserQuestion.

Batching Rule: AskUserQuestion supports max 4 questions per call.

• Batch 1: CRITICAL and HIGH severity findings (up to 4)
• Batch 2: Remaining HIGH and MEDIUM findings (up to 4)
• Continue until all findings are addressed

For each finding, create a question:

AskUserQuestion(questions=[
    {
        "question": "[finding_id]: [finding_title]\nSeverity: [severity] | How should we fix this?",
        "header": "[finding_id]",  # Max 12 chars
        "multiSelect": false,
        "options": [
            {
                "label": "[option_a_label]",
                "description": "[option_a_description - first 100 chars]"
            },
            {
                "label": "[option_b_label]",
                "description": "[option_b_description - first 100 chars]"
            },
            # ... up to 3 options
        ]
    },
    # ... up to 4 questions per batch
])

The "Other" option is automatically included by AskUserQuestion.

After each batch, if more findings remain, call AskUserQuestion again.

Step 6: Generate Implementation Summary

Based on user selections, generate an expert end-user summary.

Format:

# Red Team Fixes - Selected Actions

## Summary
[N] issues addressed | Touches: [list of affected components]

---

## [finding_id]: [finding_title]

**Issue**: [Brief description of the problem and its risk]

**Selected fix**: [Selected option label]

**What changes**:
- [Change 1]
- [Change 2]
- [Change 3]

**Why this over alternatives**:
[Brief explanation of why this option was selected over others]

**Watch out for**: [Any risks or things to test]

---

[Repeat for each selected fix]

---

## Suggested Order
1. [First thing to do]
2. [Second thing to do]
3. [Continue...]

If the user selected "Other" for any finding, include their custom input in the summary.

Step 7: Return Output

Return the implementation summary DIRECTLY to the user.

DO NOT:

• Add commentary about the process
• Explain what the agents did
• Include raw YAML output
• Show intermediate findings

ONLY return the final implementation summary.

Context Isolation Rules

This command is the BRIDGE between main session and red team work:

• Main session context stays CLEAN
• Only structured snapshot data passes to coordinator
• Fix options return as structured YAML
• Menu interaction happens in main context (safe)
• Only sanitized summary returns to user

Error Handling

If the coordinator returns no findings:

# Red Team Analysis Complete

No issues at CRITICAL, HIGH, or MEDIUM severity were identified.

The analysis found [N] LOW/INFO level observations which don't require fixes.
Run `/redteam` for the full report if interested.

If a batch of questions returns with all "skip" or empty responses:

• Continue to next batch
• Note skipped findings in summary

Example Flow

1. User runs: /redteam-w-fix standard
2. Command builds snapshot, launches coordinator
3. Coordinator returns 3 findings with fix options
4. Command presents menu:

   [RF-001] [AG-002] [CM-003]
   
   RF-001: Invalid inference in auth
   Severity: HIGH | How should we fix this?
   
   ○ A: Add validation [LOW]
   ○ B: Refactor flow [MEDIUM]
   ○ C: Type-safe handlers [HIGH]
   ○ Other
   

5. User selects B for RF-001, A for AG-002, Other ("Just add logging") for CM-003
6. Command generates summary with all selections
7. User receives actionable implementation summary