groundwork
Establish the missing engineering baseline for a repository's purpose and prove first green. Use for a request to set a repo up properly or bring it to a working baseline; preserve established components and avoid unnecessary infrastructure.
- License 许可
- MIT
- 最近更新
- 2026-10-21
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- groundwork
- source
- greenline native setup workflow; the work guide and the ledger guide own its integration.
- description
- Establish the missing engineering baseline for a repository's purpose and prove first green. Use for a request to set a repo up properly or bring it to a working baseline; preserve established components and avoid unnecessary infrastructure.
02 资源文件 1
- 参考文档references/greenline.md1.5 KB
03 SKILL.md 正文
Groundwork
Groundwork is the start for a setup request: set this repository up properly,
or bring it to a working baseline. Read the existing repository and settled
decisions first, and establish the purpose and the choices before any edit.
Then hand the work to a mode: propose it and wait, as AGENTS.md says, open the
change record with a ## start section naming groundwork and the combination
it recommends, and run the line from frame. Compare choices only where this
component lacks an established answer. The setup request authorizes ordinary
reversible work in its scope; clarify only consequential choices the request
and evidence leave open.
On an empty tree groundwork is also the orientation AGENTS.md names: its own change, proposed before the feature a first request asks for, or that request's own change when it asks to set the repository up.
Greenline's paths and commands for this skill stand on its references page, references/greenline.md: read it when the work lands in the workspace.
Establish configuration
If initialization is part of the request, it is the change's first edit, at its build step: install the practice for the harnesses the owner works in, as groundwork's references page says. An existing initialized repository is not reset because a request fails.
Establish purpose and constraints
Identify the first real use, deployment constraints, component roots, existing tools, collaborators, and relevant environment boundaries. An established tree may need a repaired check rather than a new stack. A disposable script may need only its lightweight execution form. Recommend a coherent combination where choices are open, with its meaningful trade-offs and compatibility evidence. Ask about an unresolved license or consequential platform choice before choosing it; leave an unanswered choice open. Do not infer publication or paid services from a request for local setup. On an empty repository the toolchain is this method's to choose from the brief and the runtime: decide it, record it as the toolchain decision in the book with its rationale row, and name it in the reply in one line.
Build the missing baseline
The change's build step builds the baseline. Its plan step splits the work into tickets only where distinct outcomes or dependencies need them; The work guide says where those files live, and no fixed ticket count or extra grant artifact is required.
Assess which of these outcomes the purpose actually needs:
- A coherent language and tool combination, with current compatible versions
and its scoped rationale recorded once in the decisions book.
Write its explicit root statement using the exact
greenline-rootsformat the work guide gives under Repository memory. The rationale field isdecision. Unknown purpose stays null; unknown choices stay empty. Keep siblings separate. - A reproducible check that exercises the actual code and relevant failure paths. Use the repository's dependency resolution and execution conventions.
- Appropriate ignores, editor settings, and environment boundaries where those are needed. Secret values belong outside repository evidence.
- A clear starting README and the owner's licensing choice.
- Domain vocabulary or architectural decisions when there is something durable to name or explain; empty documents do not improve the baseline.
Preserve established choices outside the requested change. Resolve volatile compatibility from primary sources when it matters. Choose only roles the component needs; an option is not an obligation to install a dependency or invent a service.
Prove first green
Run the actual pinned tools and frozen dependency resolution as the check will use them, each recorded as a check, so the change's record keeps the facts of every run. Preserve failed runs and fix their causes. Commit a coherent result, each commit carrying the change's trailer, for the review step at the depth its mode gives, then verify the change's acceptance against that result at the verify step's depth. Shared measurements are cited, not represented as separate runs.
Reconcile relevant changes to repository choices in the same work. Report what is usable, the important choices and evidence, and any open owner decision. A remote-less repository can finish with a local check. Forge protection, secrets, hosting, and publication require their own authority when needed; prepare a concrete next action without turning them into setup busywork.
Handoff
Consumes: the repository as found and its settled decisions Produces: the recommended combination and the owner's open choices, in the reply and the change record's start section; through the change, the toolchain, check, ignore and README baseline, a rationale row and an explicit greenline-roots root statement in .greenline/DECISIONS.md Next: frame, under the mode the owner approves
04 同仓库更多 Skills
use-garden
Consulting garden, the separately installed guidance command this repository's owner enabled, when an open engineering choice the repository does not settle, made in a change or only recommended, would be informed by published guidance. Use for a choice between approaches, a pattern or pitfall for a technology or concern the work touches, or a unit a repository decision names; not for a question about this repository's own facts or a change whose choices are already settled.
use-garden
Consulting garden, the separately installed guidance command this repository's owner enabled, when an open engineering choice the repository does not settle, made in a change or only recommended, would be informed by published guidance. Use for a choice between approaches, a pattern or pitfall for a technology or concern the work touches, or a unit a repository decision names; not for a question about this repository's own facts or a change whose choices are already settled.
what-to-build
Propose what this product could do next at three altitudes: sharpen a shipped slice, extend one, or open ground the product does not stand on yet. Use for 'what should we build next', 'any ideas for this app', 'what's missing', 'is this done or is there more in it', and quick-win hunts. Not for refactors, which belong to the architecture survey.
wayfinder
Plan a huge chunk of work (more than one agent session can hold) as a shared map of decision tickets in a plan directory's map.md, and resolve them one at a time until the way to the destination is clear.
verify
Prove a change's claims with fresh evidence: restate each item of its acceptance so it can fail, capture a baseline and a treatment with the same command, compare the raw artifacts, and return VERIFIED, NOT VERIFIED or INCONCLUSIVE for each, at the depth the change's mode gives the verify step. Use when the mode reaches verify, or when the owner says "verify this", "prove it works" or "did this fix it".
tdd
Test-first development at any step: discover the repository's own test commands, name the break a test catches, watch it fail for the intended reason, make it pass with the smallest change, and add a new test only where it catches a realistic failure the existing ones would miss. Use when writing or changing a test, building a feature or fixing a bug test-first, or deciding whether a change needs a new test; also on 'red-green-refactor' and integration tests.