AGENT SKILL

review-lenses

The review lenses, the checklists the reviewer applies to a diff: code, security, ui-design, ux, architecture. Reach for it when reviewing a PR, when writing a brief's `## Lenses` section, or when adding a lens to the set.

License 许可
MIT
最近更新
2026-11-15
Stars
★ 5
Forks
1

01 前言 Front Matter

name
review-lenses
description
The review lenses, the checklists the reviewer applies to a diff: code, security, ui-design, ux, architecture. Reach for it when reviewing a PR, when writing a brief's `## Lenses` section, or when adding a lens to the set.

02 资源文件 4

03 SKILL.md 正文

Review lenses

A lens is a checklist one reader applies to a diff. It is not a pass and not a ceremony: the merge gate stays one, with one verdict line and one door class, as agents/reviewer.md states them. A lens may have a specialist reader (architect, security, ux); the reader owns the lens, and the reviewer keeps the verdict, the merge and the escalation.

The refiner picks the lenses for a change and writes a ## Lenses section in the brief, after ## What: the lens names from the table below, one clause each on why. The choice comes from the change's surface, not the ticket's title. A brief with no ## Lenses is reviewed with code alone.

The floor

Every brief names the floor: code, security and architecture, one clause each on why. ui-design and ux are chosen when the change's surface earns them. This is the one home of the floor; agents/refiner.md and agents/reviewer.md point here and do not restate the list.

A lens is named either way. A named lens with no target in the change says nothing, and the brief records that. A specialist is dispatched only when its lens has a target; when it cannot be dispatched, the brief names the fallback and the pass owner applies the lens. Never a silent skip.

The lenses

lens file owner what it reads
code agents/reviewer.md reviewer the gate's own reading: correctness, edges, breakage
security skills/review-lenses/security.md security secrets and tokens, shell and SQL interpolation, input that reaches a path, writes that lose or expose store data, --force and permission changes
ui-design skills/review-lenses/ui-design.md reviewer the board's visual language: rust/core/src/serve/static/tokens.css tokens instead of literals, spacing, type, colour, dark mode, popover and sheet sizing
ux skills/review-lenses/ux.md ux interaction and recovery: keyboard paths, destructive acts without undo or confirmation, empty and error states, the human's round-trips
architecture skills/review-lenses/architecture.md architect layering, one truth per fact, schema and migration discipline, the ADRs the change touches, the reader-first order

This table is the single source of truth: a lens is a row plus a file and an owner, and the three cannot drift. code has no file of its own because agents/reviewer.md already carries that reading, and a copy would be a second contract to keep in step.

What a lens may do

Every lens file has the same three headings:

  • ## Read: what in the diff the lens looks at.
  • ## Blocks on: the findings that must change first, a REQUEST_CHANGES review, per agents/reviewer.md Commenting.
  • ## Never blocks on: the taste calls, a body comment and merge, per "Nits do not veto".

Every ## Blocks on entry cites a repo path or an ADR, never generic advice, and cites file:line in the finding itself. A lens with no basis for a judgement says nothing rather than guessing: ui-design and ux read the SwiftUI source and any screenshot the PR attaches, and judge no screen they cannot see.

A lens whose file cannot be read is named in the review body, and the gate falls back to code: a missing lens is never a silent pass.

Adding a lens

One file and one row with its owner, in the same PR, under the three headings above. A file with no row, a row with no file, and an owner with no prompt are all drift, and a reviewer checks them against this table.

04 同仓库更多 Skills

gianniacquisto
gianniacquisto/tandem

writing-for-agents

Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.

★ 5
gianniacquisto
gianniacquisto/tandem

tdd

Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.

★ 5
gianniacquisto
gianniacquisto/tandem

tandem-tidy

Board hygiene pass: sweeps the open tickets against work that already landed, completes what is done or moot with the evidence in a comment, reports stale claims, archives the done, and reports the oddities. Reach for it when the human says tidy or clean up the board, asks what on the board is stale, or right after a run of merges.

★ 5
gianniacquisto
gianniacquisto/tandem

tandem-team

The roster, which agent fills which role, what to write in `assignee` and `author`, and the minutes rule for the ticket comment log. Reach it when you need to know who owns a pass, what to write as a ticket's assignee or a comment's author, or who the team is.

★ 5
gianniacquisto
gianniacquisto/tandem

tandem-sprint

Triage, build, review and close a batch of tickets in one invocation: orient on the board, draw ready tickets only, run each through `tandem-build`, and settle the lane at the close. Refine and retro are standalone passes: shaping is `skills/tandem-refine/`, learning is `skills/retro/`. Reach for it when the user says run a sprint, asks what the workflow is, or starts a coding pass on this board.

★ 5
gianniacquisto
gianniacquisto/tandem

tandem-refine

Dispatch the refiner to shape a backlog ticket into a buildable brief, the front door for a refine pass, where the human wants to spend time refining rather than building. Reach for it when the user says to refine, spec, or shape a ticket, or asks what needs refining.

★ 5