superreview

opus generator human-invoked

Note

Generator skill. /brewcode:superreview does not review code itself. It analyzes the target project and writes a self-contained .claude/skills/superreview/ into it. The emitted skill does the reviewing.

Quick reference

Field	Value
Command	`/brewcode:superreview "<fine-tune-prompt>" [scope]`
Arguments	`<fine-tune-prompt>` — free-text priority emphasis woven into the emitted skill; optional `[scope]` hint passed through
Model	opus
Tools	Read, Write, Edit, Bash, Task, AskUserQuestion
Output	`.claude/skills/superreview/` in the target project

What it does

superreview is a generator skill. It scans the target project — enumerating agents, rule files, source directories, and tech stack — then writes a self-contained review skill wired to that project’s actual structure. No plugin dependency remains after generation.

The emitted skill fuses two earlier passes: the review engine (two-phase find-then-validate, per-finding adversarial gate, single merged P0-P3 report) and the standards-review pass (reuse and duplication focus, 90/70/50 percent reuse matrix, per-stack guidelines). Both run as one /superreview invocation, producing one report.

The <fine-tune-prompt> argument shapes the emitted skill’s Focus rank table. Values such as “treat any security issue as P0” or “focus on architecture boundaries and reuse” are woven into the emitted SKILL.md so every /superreview run in the target project reflects those priorities.

When to use

Situation	Command
Wire up a new project for deep review	`/brewcode:superreview "default ordering"`
Security is the top concern	`/brewcode:superreview "treat any security issue as P0"`
Architecture and reuse are the priority	`/brewcode:superreview "focus on architecture boundaries and reuse"`
An agent, rule file, or source group changed	Re-run to refresh routing and rule pointers

Example

/brewcode:superreview "focus on reuse and duplicate-detection"

The generator scans the project, asks about ambiguous domain mappings if needed, then writes:

superreview generated -> .claude/skills/superreview/

Stack:         Node/TypeScript  (reference: typescript-react.md)
Domain agents wired: 3  (developer, tester, architect)
Review agents: reviewer, architect
Rule pointers: 5 files referenced (not restated inline)
File groups:   4  (src/developer, tests/tester, infra/architect, docs/general)
Focus:         1 reuse  2 correctness  3 architecture  4 version-pins  5 business-reqs
               fine-tune: "focus on reuse and duplicate-detection"

Files written:
  .claude/skills/superreview/SKILL.md
  .claude/skills/superreview/references/agent-prompt.md
  .claude/skills/superreview/references/report-template.md
  .claude/skills/superreview/references/typescript-react.md

Run it:  /superreview "<focus>" [scope: commit|branch|folder]

To run the emitted skill after generation:

/superreview "focus on reuse" last_3

Workflow

Phase 0 — Load templates
The generator reads the four emit templates bundled with the skill: SKILL.md.template, agent-prompt.md, report-template.md, and the stack reference files. These are the scaffold for the emitted skill.
Phase 1 — Scan the target project
Runs generate.sh scan to enumerate agents, rule files, source directories, and tech stack. Produces a SCAN_REPORT with agent list, rule list, source groups, and stack hint.
Phase 1.5 — Clarify ambiguous params
Sends an AskUserQuestion when stack detection is ambiguous, domain-agent mapping is uncertain, or source group boundaries are unclear. Skipped when all params resolve unambiguously from the scan.
Phase 2 — Scalar substitution
Calls generate.sh emit to fill all deterministic placeholder values: agent names, rule file paths, file groups, stack reference path. Writes initial file content to .claude/skills/superreview/.
Phase 3 — AI fills block placeholders
The AI edits the emitted files via the Edit tool to fill block sections: per-domain review rules, focus rank table, tech-stack check list, and adversarial validation prompt.
Phase 4 — Validate
Runs generate.sh validate to confirm no unresolved placeholder tokens remain. Errors block completion and prompt a targeted Phase 3 retry.
Phase 5 — Report
Prints a summary: files written, agent wiring table, focus rank, and the invocation command for the emitted skill.

Technical details

Emitted skill — review modes

The emitted /superreview is fully self-contained after generation. It resolves the review mode from the scope argument:

Mode	Scope keyword
`FULL_PROJECT`	No scope, or explicit `full`
`EXPLICIT`	Named files or directories
`UNCOMMITTED`	`uncommitted` / `staged`
`LAST_COMMITS`	`last_N` / `branch`

The emitted skill is READ-ONLY by design — it reports findings, never edits source code. Per-finding adversarial validation: for each finding in the parallel fan-out, a second agent pass challenges it before it enters the merged report. Findings that fail validation are dropped or downgraded. Output: .claude/reports/{TIMESTAMP}_superreview/REPORT.md.

Re-run triggers

Regenerate the emitted skill when any of the following change in the target project:

Agent added, removed, or renamed
Rule or convention file added
Tech stack changes (new runtime, framework, or language)
Source group boundaries shift (new top-level directory)

Error handling

Error	Resolution
No agents found in target project	Generator stops; create at least one agent first
Unknown tech stack	Emits with generic reference; warns to review manually
Multi-stack project	Emits one reference per detected stack; adds merge note
Unresolved placeholder after Phase 3	Phase 4 blocks completion; re-run Phase 3 on affected file
Unwritable target path	Generator stops; check permissions on `.claude/skills/`

Emitted file structure

.claude/skills/superreview/
  SKILL.md                          # emitted review skill (self-contained)
  references/
    agent-prompt.md                 # per-agent review instructions
    report-template.md              # merged P0-P3 report template
    {stack}.md                      # java-kotlin | typescript-react | python | go

📄

Updating plugins

Use /brewtools:plugin-update to check and update the brewcode plugin suite in one command. See the FAQ for details.