/unified-review

name: unified-review
description: Orchestrates multi-domain review (code, arch, tests, security) in a single pass. Use when thorough pre-release review is needed.
alwaysApply: false
category: orchestration
tags:
- review
- orchestration
- code-quality
- analysis
- multi-domain
tools: []
usage_patterns:
- auto-detect-review
- full-review
- focused-review
complexity: intermediate
estimated_tokens: 400
progressive_loading: true
dependencies:
- imbue:proof-of-work
- imbue:review-core
- imbue:structured-output
modules:
- modules/review-workflow-core.md
- modules/output-format-templates.md
- modules/quality-checklist-patterns.md
- modules/shell-review.md
- modules/shell-exit-codes.md
- modules/shell-portability.md
- modules/shell-safety-patterns.md
- modules/shell-structure-patterns.md
orchestrates:
- pensive:rust-review
- pensive:api-review
- pensive:architecture-review
- pensive:bug-review
- pensive:test-review
- pensive:makefile-review
- pensive:math-review
- imbue:justify

Unified Review Orchestration

Intelligently selects and executes appropriate review skills based on codebase analysis and context.

Quick Start

# Auto-detect and run appropriate reviews
/full-review

# Focus on specific areas
/full-review api          # API surface review
/full-review architecture # Architecture review
/full-review bugs         # Bug hunting
/full-review tests        # Test suite review
/full-review all          # Run all applicable skills

**Verification:** Run `pytest -v` to verify tests pass.

When To Use

• Starting a full code review
• Reviewing changes across multiple domains
• Need intelligent selection of review skills
• Want integrated reporting from multiple review types
• Before merging major feature branches

When NOT To Use

• Specific review type known
• use bug-review
• Test-review
• Architecture-only focus - use

architecture-review

• Specific review type known
• use bug-review

Review Skill Selection Matrix

| Codebase Pattern | Review Skills | Triggers | |-----------------|---------------|----------| | Rust files (`*.rs`, `Cargo.toml`) | rust-review, bug-review, api-review | Rust project detected | | API changes (`openapi.yaml`, `routes/`) | api-review, architecture-review | Public API surfaces | | Test files (`test_*.py`, `*_test.go`) | test-review, bug-review | Test infrastructure | | Makefile/build system | makefile-review, architecture-review | Build complexity | | Shell scripts (`*.sh`, git hooks) | shell-review module | Shell changes | | Mathematical algorithms | math-review, bug-review | Numerical computation | | Architecture docs/ADRs | architecture-review, api-review | System design | | General code quality | bug-review, test-review | Default review | | Post-implementation audit | imbue:justify | High add/delete ratio, test changes, new abstractions |

Workflow

1. Analyze Repository Context

• Detect primary languages from extensions and manifests
• Analyze git status and diffs for change scope
• Identify project structure (monorepo, microservices, library)
• Detect build systems, testing frameworks, documentation

2. Select Review Skills

# Detection logic
if has_rust_files():
    schedule_skill("rust-review")
if has_api_changes():
    schedule_skill("api-review")
if has_test_files():
    schedule_skill("test-review")
if has_makefiles():
    schedule_skill("makefile-review")
if has_shell_scripts():
    schedule_module("shell-review")
if has_math_code():
    schedule_skill("math-review")
if has_architecture_changes():
    schedule_skill("architecture-review")
# Default
schedule_skill("bug-review")

**Verification:** Run `pytest -v` to verify tests pass.

3. Execute Reviews

Dispatch selected skills concurrently via the Agent tool. Use this mapping to resolve skill names to agent types:

| Skill Name | Agent Type | Notes | |---|---|---| | bug-review | `pensive:code-reviewer` | Covers bugs, API, tests | | api-review | `pensive:code-reviewer` | Same agent, API focus | | test-review | `pensive:code-reviewer` | Same agent, test focus | | architecture-review | `pensive:architecture-reviewer` | ADR compliance | | rust-review | `pensive:rust-auditor` | Rust-specific | | code-refinement | `pensive:code-refiner` | Duplication, quality | | math-review | `general-purpose` | Prompt: invoke `Skill(pensive:math-review)` | | makefile-review | `general-purpose` | Prompt: invoke `Skill(pensive:makefile-review)` | | shell-review | `general-purpose` | Prompt: follow this skill's `modules/shell-review.md` |

**Sub-agent isolation (required).** One lens must not color how the next is read. Two ways to get that, and the second only runs when the user asks for it:

| Path | How isolation holds | |------|---------------------| | Agent tool | Dispatch ALL selected agents in a SINGLE parallel call, and read no output until every agent has returned. Reading the first result anchors synthesis toward it: each later result gets judged against the first rather than independently. Collect all, then synthesize once | | `/pensive:unified-review` workflow (`workflows/unified-review.js`) | The script holds the results. It is not a reasoning entity, so it cannot be anchored by reading one stage before another, and findings pass between stages without entering anyone's context. Verification also starts per dimension instead of waiting for the slowest lens |

Prefer the workflow when the dimension list is known before the work and each finding should face an adversarial check. Prefer the Agent tool when the roster has to adapt to what the first lens finds, or when no workflow was requested: a workflow never starts unasked.

Its subagents take the session's permission mode, so in an accept-edits or auto session their edits land unreviewed. The shipped script scopes its prompts to reading. And it has no filesystem, so what it returns is a claim that survived refutation, not proof-of-work evidence. Reproduce before acting on a finding.

**Rules:**

• Never