Skip to content
Skills Library
Programming · Testing

select

/select

Use when a code change needs its E2E scope decided — before running Playwright on a branch, worktree or PR, or when asked "which E2E specs does this change need?". Computes the selection with the e2e-rail CLI, checks the couplings the script cannot see, and produces the selection block (change · selected specs · reasons · mobile · unmapped · final full) that every selected-run report must carry. Do NOT load for: running or reporting tests (gate), first-time setup (init), speed, workers or sharding (measure), shadow pairing or promotion (shadow).

Repository
e2e-rail ↗
Licence
MIT
Last repository change
27.10.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 27.10.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Use when a code change needs its E2E scope decided — before running Playwright on a branch, worktree or PR, or when asked "which E2E specs does this change need?". Computes the selection with the e2e-rail CLI, checks the couplings the script cannot see, and produces the selection block (change · selected specs · reasons · mobile · unmapped · final full) that every selected-run report must carry. Do NOT load for: running or reporting tests (gate), first-time setup (init), speed, workers or sharding (measure), shadow pairing or promotion (shadow).

Triggers from the description

which E2E specs does this change need?
Original author description

Use when a code change needs its E2E scope decided — before running Playwright on a branch, worktree or PR, or when asked "which E2E specs does this change need?". Computes the selection with the e2e-rail CLI, checks the couplings the script cannot see, and produces the selection block (change · selected specs · reasons · mobile · unmapped · final full) that every selected-run report must carry. Do NOT load for: running or reporting tests (gate), first-time setup (init), speed, workers or sharding (measure), shadow pairing or promotion (shadow).

How it works

  1. 1. Pick the base
  2. 2. Compute
  3. 3. Check what the script cannot see (partial apps only)
  4. 4. Write the selection block (paste it into your report verbatim)
  5. 5. Hand off
Original SKILL.md headings · 92 lines, 1,035 words
SKILL.md preview
---
name: select
description: Use when a code change needs its E2E scope decided — before running Playwright on a branch, worktree or PR, or when asked "which E2E specs does this change need?". Computes the selection with the e2e-rail CLI, checks the couplings the script cannot see, and produces the selection block (change · selected specs · reasons · mobile · unmapped · final full) that every selected-run report must carry. Do NOT load for: running or reporting tests (gate), first-time setup (init), speed, workers or sharding (measure), shadow pairing or promotion (shadow).
---

# e2e-rail:select — decide the E2E scope and record why

Run every command from the repository root. The CLI: `e2e-rail` (Claude Code puts the plugin's copy on PATH; it runs the copy the repository pins in
`node_modules` when there is one). Not on PATH (Codex, CI): `pnpm exec e2e-rail`, or `npx e2e-rail` in npm and yarn
repositories. Without `--app`, `select` covers every app of the config; `--add`, `run` and
`verify` need `--app <app>` when there are several.

## 1. Pick the base
The base is the last commit that passed a full verification on a clean tree. In this order:
1. `e2e-rail verify --app <app> --mode <mode>` — `verified:` means this exact code already passed in full:
   nothing to select, report it through `e2e-rail:gate`. A `stale:` line that ends with `last verified head <sha>`:
   use it. `no full pass of a committed tree yet` (every pass had uncommitted changes), or
   `run <id> was recorded under an older e2e-rail verification policy` (only an older e2e-rail passed it): go on to 2.
2. `--base last-green` — e2e-rail reads it from the ledger: the head of the app's last unfiltered passing full run, or
   `shard merge` set that printed `complete: yes`, of a clean tree, recorded under the current verification policy;
   never a pass with uncommitted changes. With none it is no base (`no base (last-green: …)`, full). Do not read
   `.e2e-rail/last-green.<app>` yourself: that file is informational, and one an older e2e-rail wrote may name a
   dirty or relaxed pass. `--base last-green` takes the newest clean full pass in ANY mode (the header says which:
   `base <sha7> (last-green: run <id>, <mode>)`); to diff from a commit verified in preview, pass that commit's SHA.
   `last-green` is a keyword: pass a branch or tag literally named `last-green` as its SHA.
3. The merge-base with the target branch: `git merge-base HEAD origin/main`.

No base → leave `--base` out and accept a full selection. Never pick a base to get a smaller selection.

## 2. Compute
```sh
e2e-rail select --app <app> --base <sha>   # or --base last-green
```
The selection stores the commit the base named (header line `base <sha7> (<ref>)..HEAD`), never the name: later
commits on HEAD or the branch cannot move it.
Exit 0 = every covered app is partial · 10 = some app runs in full · 1 error · 2 usage. Locally, uncommitted and
untracked files are included (`+ uncommitted` in the header line); `--no-uncommitted` leaves them out; under `CI` they
are out by default — on a tree that has any, every app then runs full (`uncommitted-excluded`). Output: the table `app  mode  specs  unmapped  reasons`, then `.e2e-rail/selection.json` (per app:
`mode`, `specs[].file`, `specs[].projects`, `specs[].reasons`, `changedFiles`) and, for a partial app,
`.e2e-rail/test-list.<app>.txt` (lines `[<project>] › <spec path from Playwright's rootDir>`).

Reasons on a selected spec: `spec-self:<file>` · `route:<route> ← <file>` · `api:<glob> ← <literal>` ·
`import:<file>` · `unmapped` (the index cannot map that spec, so it runs on every src change) · `always-run` ·
`added: <reason>`.

A full app names why. Whatever the script cannot attribute widens, never narrows: `no-base` ·
`head-not-HEAD:<ref>` (`--head` other than HEAD: a run tests the work tree) · `uncommitted-excluded` (uncommitted work
left out, yet the run tests it) · `shared:<file>` ·
`unknown-root:<file>` (outside every app root or outside the config root: every app runs in full) ·
`tier-full:<glob>` · `support:<file>` · `app-other:<file>` · `spec-unindexed:<file>` · `adapter-unresolved:<…>` ·
`graph-unresolved:<file>` (also when any internal import does not resolve) · `graph-shell:<file>` (the change
climbs to the app entry: shell, stores, layout). A full app is a correct answer; do not argue it down.

## 3. Check what the script cannot see (partial apps only)
- Claude Code: delegate to the `e2e-impact-analyst` agent with the diff (`git diff <base>` plus `git status --short`)
  and the selection file (the `selection:` line of `select` prints its path: `.e2e-rail/` in project mode,
  `~/.e2e-rail/<repo>/ledger/<worktree>/` in personal mode).
- Codex: read `references/impact-analyst.md` of this plugin (`../../references/impact-analyst.md` from this file)
  and run its checklist yourself.

It returns `- add e2e/<spec>.spec.ts — <reason>` lines or `no additions`. Apply each addition:
```sh
e2e-rail select --app <app> --add e2e/<spec>.spec.ts --reason "<one sentence>"
```
`--add` amends the current selection and rewrites its test list; it does not take `--base`/`--head`. Spec paths are
app-relative, as `map --explain` prints them. One call per addition keeps each reason (several `--add` in one call
share one `--reason`). After the last addition, record the check (a partial selection only):
```sh
e2e-rail select --review-done --reason "impact-analyst: <no additions | n additions>"
```
It binds the check to this code and this list: a later `--add`, or any code change, makes it stale, and
`run --selection` and `status` print `review: …`. The record says the check was made, not that it found everything.

Never `--remove` while `e2e-rail shadow status` prints `trust shadow` (the CLI refuses it anyway). After a
human ran `shadow promote`: `select --remove <spec> --reason "<why>"`, and the removal goes into your report.

## 4. Write the selection block (paste it into your report verbatim)
```
change:   <what changed, one line> (<top files> +<n> more)
selected: <spec slugs> (reasons: route:… · api:… · import:… · unmapped · always-run)
added:    <slug — reason> | none      removed: <slug — reason> | none
mobile:   <slugs whose test-list lines include a mobile project (a project with a mobile device)> | none
review:   <the review line run --selection prints: ok (<reason>) | none | stale (<why>) | not needed>
unmapped: <the unmapped column of the select table> (map --check lists them)
final:    full run in the integration step · trust <shadow|selected> (as shadow status prints it)
```
For an app that runs in full, write `selected: all — app runs in full (<reasons>)` and `mobile: all projects`.
One block per app.

## 5. Hand off
`e2e-rail:gate` runs it (`run --selection`). Do not run `playwright test` yourself. If files change after this step,
`run --selection` selects again from the same base commit (`selection <old> was for other code — reselected as <new>`)
and keeps your `--add`s, not your `--remove`s: update the block from the new selection. A selection an older e2e-rail
wrote kept its base by name; it is reselected in full
(`selection <old> stored its base by name ("<ref>"), which may have moved — reselected in full`).

Versions

RepositoryScoreLicence
This page · sh5623/e2e-rail100MIT
rhiza-research/weather-skills100MIT

Identical content appears in 1 other GitHub folders. We show one copy with the highest score.

Database snapshot: 13.11.2026 · package version b653aaf73794 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100init
    Use when a repository with Playwright E2E tests does not use e2e-rail yet (no e2e-rail.config.mjs, or `e2e-rail` is missing), or when its config needs review after adding an app, a route table, a monorepo package or a new tsconfig. Installs the CLI, writes the config, checks how…
  • 100shadow
    Use when a round ends, after the integration step's full E2E run, or when asked whether selected runs can be trusted to replace full runs. Pairs the full run with the selection computed for the same code, records hit or miss with the e2e-rail CLI, reports the streak and puts…
  • 100playwright-regression-testing
    Govern Playwright TypeScript regression suites across many tests. Use when asked to plan, select, tier, execute, or optimize suites with risk/change analysis, tags, CI/CD, sharding, flaky-test management, or suite-health metrics; not for authoring one UI spec. Keywords:…
  • 100pw-init
    Set up Playwright in a project. Use when user says "set up playwright", "add e2e tests", "configure playwright", "testing setup", "init playwright", or "add test infrastructure".
  • 100pw-init
    Set up Playwright in a project. Use when user says "set up playwright", "add e2e tests", "configure playwright", "testing setup", "init playwright", or "add test infrastructure".
  • 100pw-init
    Set up Playwright in a project. Use when user says \"set up playwright\", \"add e2e tests\", \"configure playwright\", \"testing setup\", \"init playwright\", or \"add test infrastructure\".

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.