Apply structured strategic thinking before implementation. Use when facing architectural decisions, system design, refactoring, build-vs-buy choices, debugging complex issues, planning migrations, writing technical proposals, or any task where jumping straight to code would be premature. Triggers on: "design", "architect", "plan", "strategy", "evaluate", "decide", "trade-off", "should we", "how should", "what approach", "refactor", team dynamics questions, or when the problem has multiple valid solutions. Also activates on /strategize.
Test semantics audit. Distinguishes tests that prove business rules from tests that merely exercise code paths for coverage. Applies Islamic epistemology cognitive levels (Hissi through Fikri) to classify test proof depth. Auto-triggers on: /istishraf:testaudit, "are my tests fake", "coverage theater", "do my tests prove", "test quality", "are tests meaningful", "business logic coverage", "test intent", "mutation resistant tests", "tests just for coverage", or any question about whether a test suite actually validates expected behaviour vs. just passing.
Adversarial code review skill. Attack implementation before it reaches production. Applies 7 attack vectors: concurrency (TOCTOU), input validation, state assumptions, time-of-check vulnerabilities, security (injection, XSS, privilege), transaction isolation, and the 3am production test. Auto-triggers on: /review, "review this code", "check for issues", "is this safe", "attack this", or at Phase 7 of /implement.
Extended cognitive analysis combining multiple reasoning frameworks. Use when /think-deeper is invoked, when strategic analysis yields conflicting signals, when the problem has ethical or philosophical dimensions, when team dynamics are central, or when the user asks for "multiple perspectives", "deeper analysis", "six hats", "cognitive levels", "bridges", "ibn khaldun", or "wisdom perspective."
Disciplined 8-phase implementation methodology. Applies architect-mode execution: codebase-first exploration, test-driven development (red-green- refactor), adversarial self-review, and automated quality gate cycle. Auto-triggers on: /implement, "let's build", "write the code", "implement", "fix the bug", "add the feature", or any task where the strategy is decided and execution begins. Also activates when the user pastes code and asks for changes to multi-file systems.
Multi-perspective code review council. Convenes distinct reviewer personas (Architect, Adversary, Maintainer, Operator, Product Guardian, Security Auditor) to examine code from different professional angles. Auto-triggers on: /shura, "review from multiple perspectives", "different angles", "another role", "council review", "review as a", "what would a [role] think", "security review", "architectural review of my changes", or when the user wants richer feedback than a single adversarial pass.