Review
Structured, multi-lens audit of a project — docs, UX, marketing, or product. Reads current state, finds gaps, proposes concrete fixes. Invoke as /flow:review [--docs|--ux|--marketing|--product]; runs all four lenses when none is named.From its SKILL.md
npx -y skills add liminalarc/flow-toolkit --skill reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.4 KB, as published. Nobody here has run it
flow:review
Structured audit from multiple lenses. Reads current state, identifies gaps, proposes concrete improvements. Run all lenses or narrow to one.
Usage:
/flow:review— all lenses/flow:review --docs— documentation only/flow:review --ux— UX audit only/flow:review --marketing— marketing/positioning only/flow:review --product— product critique only
Instructions
Start fresh. Read only from the project files — CLAUDE.md, SPECIFICATIONS.md, README.md, MARKETING.md. Do not reference or build on prior conversation context. Treat this as a new session regardless of what preceded it.
Fan out to one reviewer per lens. Each lens is audited by an independent, read-only flow:flow-reviewer sub-agent that loads only its own rubric — so the main thread stays lean and lenses run in parallel. Lens → rubric:
| Lens | Flag | Rubric (passed to the reviewer) |
|---|---|---|
| Docs | --docs | reference/docs.md |
| UX | --ux | reference/ux.md |
| Marketing | --marketing | reference/marketing.md |
| Product | --product | reference/product.md |
Dispatch. For each requested lens, launch a flow:flow-reviewer agent with: the lens name, the path to its rubric (reference/<lens>.md within this skill's directory), and the project root. When more than one lens runs, launch them in parallel — a single message with multiple agent calls. Each reviewer returns prioritized findings for its lens and never edits.
Synthesize (main thread).
- Single lens → present that lens's prioritized findings.
- All lenses (
/flow:review, no flag) → run all four reviewers, then produce a cross-lens summary of the highest-priority items across docs, UX, marketing, and product.
Applying fixes. Reviewers only report. When a finding warrants an edit (e.g. a doc fix), you — the main thread — propose it and confirm before making significant changes. The review never edits silently.
Rules
- Read before writing — understand current state before proposing changes.
- Confirm before making significant doc changes.
- Actionable findings only — specific suggested fixes, not just observations.