red-team

Red-team an implementation plan — dispatches challenger agents to find blind spots, failure modes, and scope issues

Use this agent to analyze task dependencies and ordering in implementation plans. Examines sequencing, identifies hidden dependencies between tasks, and finds parallelism opportunities. Examples: <example> Context: A plan has 12 sequential tasks but some look independent. User: "Here's my plan for setting up the monitoring stack." Assistant: "Let me use the dependency-challenger to analyze task ordering and find hidden dependencies." <Task tool invocation to launch dependency-challenger agent> </example> <example> Context: A refactoring plan touches shared modules. User: "Check my plan for splitting the monolith into separate packages." Assistant: "Let me use the dependency-challenger to trace which tasks depend on shared code." <Task tool invocation to launch dependency-challenger agent> </example>

Use this agent to find gaps in implementation plans. Looks for missing steps — untested edge cases, unhandled error paths, missing rollback procedures, missing config changes, and missing documentation. Examples: <example> Context: A plan for deploying a new service with no mention of rollback. User: "Here's my plan for deploying the new payment service." Assistant: "Let me use the gap-analyzer agent to check for missing steps." <Task tool invocation to launch gap-analyzer agent> </example> <example> Context: A plan for adding authentication that doesn't mention existing endpoints. User: "Check my plan for adding JWT auth to the API." Assistant: "Let me use the gap-analyzer to find gaps — like existing endpoints that need auth added." <Task tool invocation to launch gap-analyzer agent> </example>

Use this agent to find risks in implementation plans. Combines assumption analysis with failure pre-mortem — identifies what the plan takes for granted, then imagines failure and traces root causes backward. Examples: <example> Context: A plan proposes adding a new API integration. User: "Here's my plan to integrate the Stripe API for payments." Assistant: "Let me use the risk-analyst agent to surface unstated assumptions and likely failure modes." <Task tool invocation to launch risk-analyst agent> </example> <example> Context: A plan for a database migration. User: "Review my plan for migrating from PostgreSQL 14 to 16." Assistant: "I'll run the risk-analyst to identify what this plan takes for granted and where it's most likely to fail." <Task tool invocation to launch risk-analyst agent> </example>

Use this agent to challenge scope in implementation plans. Flags YAGNI violations, over-engineering, unnecessary complexity, and tasks that could be deferred or removed entirely. Examples: <example> Context: A plan includes building a plugin system for a feature with one use case. User: "Here's my plan for the new notification system with a plugin architecture." Assistant: "Let me use the scope-challenger to evaluate whether the plugin architecture is justified." <Task tool invocation to launch scope-challenger agent> </example> <example> Context: A plan has 15 tasks for what seems like a 5-task feature. User: "Review my implementation plan for adding dark mode support." Assistant: "I'll run the scope-challenger to identify which tasks are essential for v1." <Task tool invocation to launch scope-challenger agent> </example>