Skip to content
Skills Library
Programming · Building & coding

stacks

/stacks

Start a new project from a stack or add fragments to an existing repo, then drive it to a verified working state. Use for project setup, tests, lint, CI and agent configuration, or when the user mentions stacks.

Repository
stacks ↗
Licence
Apache-2.0
Last repository change
08.11.2026

Assessment and security

95/ 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.
Content30 / 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 08.11.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Start a new project from a stack or add fragments to an existing repo, then drive it to a verified working state. Use for project setup, tests, lint, CI and agent configuration, or when the user mentions stacks.

Original author description

Start a new project from a stack or add fragments to an existing repo, then drive it to a verified working state. Use for project setup, tests, lint, CI and agent configuration, or when the user mentions stacks.

How it works

  1. When to use it
  2. How to run it
  3. Pick a stack
  4. Start it
  5. Loop to green
  6. Never weaken verify
  7. Finish
Original SKILL.md headings · 44 lines, 600 words
SKILL.md preview
---
name: stacks
description: Start a new project from a stack or add fragments to an existing repo, then drive it to a verified working state. Use for project setup, tests, lint, CI and agent configuration, or when the user mentions stacks.
---

# stacks

## When to use it

Use stacks for a new project, or when the user wants one set up properly. For an existing repo, use `<cli> add <fragment...> --json` to apply fragments without restructuring it. Use `<cli> list --json` to choose fragments; inside a stacks project, `project.addable` lists those that fit.

A stack is a recipe: dependencies by role, starter files, agent guidance and a definition of done. A fragment is one capability a stack can include, such as CI or a database. The CLI writes the brief; you build the project with current versions.

To write or submit a stack or fragment, run `<cli> guide author`.

## How to run it

Use `stacks` on PATH, otherwise `npx -y @steamfield/stacks@latest`; `<cli>` means that command. After `new`, use the brief's verify command, which preserves its version. `<cli> <command> --help` lists flags, exits and examples.

With `--json`, a command writes one result to stdout. On failure the last stderr line is `{ "error": { "tag", "message", "retryable", "next", "details" } }`: branch on the exit code (2 fix the input, 3 fix the config, 5 read the verify report on stdout, 6 ask the user) and do what `next` (in text output, the `Next:` line) says. `<cli> schema <name>` prints the JSON Schema of a `--json` result.

Relay a telemetry `notice` once, including `<cli> telemetry disable`; never enable it for the user. Attribute `new` and `add` with `--source skill`.

## Pick a stack

Run `<cli> search "<their words>" --json`. Use `--local` for local ranking; `--offline` and `DO_NOT_TRACK` also disable API search. Compare strong matches using `why` and `evidence`. If `noStrongMatch` is true, say nothing fits well; run `<cli> list --json` and offer the closest entries. Review candidates with `<cli> show <id> --json`; explain your pick and one tradeoff. Never pick an untrusted entry (`trusted: false`) unless the user named it; if they did, confirm before continuing.

## Start it

Run `<cli> new <stack> <dir> --json`; use `<cli> new <stack> <dir> --sandbox --json` when the user wants the fastest path to something running locally. Add `--dry-run` to preview without writing.

After `add`, read the same brief, merge the contributions parked in `.stacks/incoming/` while keeping the project's choices, then delete incoming files once merged. Verify checks the whole manifest.

Read `.stacks/BRIEF.md`. The brief is the whole task: follow it in order, and ask the user only what it tells you to ask. Update the files stacks wrote to current releases; use current docs, not remembered flags.

## Loop to green

Run `<cli> verify` from the project root with your shell tool's longest timeout. Read the failing step's name, command and output tail; fix the code or config it exercises, never the step, and run it again. Before calling it done, run `<cli> verify --all` and reach exit 0 with outcome `green`.
## Never weaken verify

Don't edit `.stacks/manifest.json`. Never edit the steps or weaken the definition of done: don't make commands pass without doing their job, skip tests, lower lint settings until they report nothing, or serve a stub instead of the app.

If verify fails with `VerifyChanged`, stop. Don't edit the steps back and forth, and don't run `<cli> verify --accept-changes` yourself: it refuses agents. Tell the user which steps or settings changed and that only they can accept the changes from their own terminal, or undo them. Then wait.

## Finish

Follow the brief's after-green list, including manual QA and checking the project documentation. Then summarize for the user: the stack and why, the decisions taken with their defaults, anything left undone, and where the brief's problems were noted (`.stacks/NOTES.md`).

Versions

RepositoryScoreLicence
This page · steamfield/stacks95Apache-2.0
randomittin/heimdall100MIT
alefiengo/stacks90MIT

Database snapshot: 13.11.2026 · package version ae9826535407 (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

  • 100stack-authoring
    How to create or revise a stack (a STACK.md recipe under packages/cli/registry/stacks) and its search card in packages/cli/registry/search/stacks so it is well named, plainly described, accurate and found by the right searches. Use it whenever someone asks for a new stack,…
  • 100v-start
    Start a new project, or bring an existing repo onto the same footing, with a proven agent harness. It sets up a CLAUDE.md contract, an AGENTS.md pointer, the PROGRESS / DECISIONS / VERIFIED / RISKS ledgers, spec and plan before code, an issue-first workflow, Conventional…
  • 100stacks-author
    Create, convert or submit a stacks entry. Use when the user wants to make a stack or fragment, turn an existing repo into a stack, write a reusable project template or recipe for coding agents, add a capability (CI, auth, a database, an app) as a fragment, write a search card,…
  • 100setup
    Use when the user wants to set up a new project or add enforcement tooling (TDD, secret scanning, file size limits, git hooks, CLAUDE.md templates) to an existing project. Also use when the user says "bootstrap", "scaffold", "set up my project", or "add quality enforcement".
  • 100scaffold
    Set up a new project so build, lint, typecheck, and test all run from day one. Use when a calibrated idea is confirmed and there is no project yet, or when an existing project has no working checks. Only invoke via the orcinus router.
  • 100setup
    Prepares a project for craft. Configures the test runner, a linter with size and complexity limits (function, class, parameters, nesting), and the `test` and `lint` scripts. Use when you start a new project, or when the project has no lint with limits or no test script yet.

Feedback and reports

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