agentleFS
IamK77IamK77/Skill/skills/engineering/groundwork/SKILL.md

groundwork

Skill2 starsChanged 31 days agoApache-2.0View on GitHub

Turn a request into an agreed, testable task definition. Use when requirements, constraints, examples, or acceptance boundaries are unclear.

9 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When this stage is fulfilled: checklist advance understand --run "$RUN".

    line 41
  2. Distinctive:

    Write concrete success and failure examples, non-goals, constraints, and interfaces affected.

    line 45
  3. Very specific:

    If an input or permission is missing, name the blocker and leave the relevant item pending.

    line 73
  4. Very specific:

    After all required work and stage closures is recorded, use checklist done --run "$RUN".

    line 79

“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- Copyright 2026 IamK77 — Licensed under the Apache License, Version 2.0. See ./LICENSE and ./NOTICE · https://www.apache.org/licenses/LICENSE-2.0 -->

groundwork

Turn a request into an agreed, testable task definition. Use when requirements, constraints, examples, or acceptance boundaries are unclear.

Inputs

The human’s request; affected users and current behaviour; known constraints; who can resolve conflicting requirements.

Run record

Use checklist 0.5 or newer for this SOP. Reading or reloading the skill must not reset work. For a new task, run checklist init "${CLAUDE_SKILL_DIR}" --new --path "/absolute/project"; retain its returned ID as RUN. For existing work, use checklist resume "$RUN", then checklist show --run "$RUN". Outside Claude, replace the skill-directory variable with this directory’s actual path.

Commands below record work after it is done. Replace evidence placeholders with concise artifact locations and observations. Manual confirmations are not independent proof; a sensor pass covers only the command and inputs it observed. verify does not close a stage; advance checks that its required readings are fulfilled. Use report --run "$RUN" for the retained history.

Procedure

understand

Restate the desired outcome and ask only the questions that change scope or acceptance. Separate facts, assumptions, and choices still owned by the human.

Record these artifacts after doing the work:

  • Users, problem, and desired outcome are stated. checklist check understand need --run "$RUN" --evidence "<artifact and observation>"
  • Material ambiguities and their decision owners are recorded. checklist check understand unknowns --run "$RUN" --evidence "<artifact and observation>"

When this stage is fulfilled: checklist advance understand --run "$RUN".

specify

Write concrete success and failure examples, non-goals, constraints, and interfaces affected. Resolve contradictions with the decision owner rather than selecting a requirement silently.

Record these artifacts after doing the work:

  • Acceptance examples include meaningful boundaries and failure behaviour. checklist check specify examples --run "$RUN" --evidence "<artifact and observation>"
  • Non-goals, constraints, and unresolved decisions are explicit. checklist check specify scope --run "$RUN" --evidence "<artifact and observation>"

When this stage is fulfilled: checklist advance specify --run "$RUN".

agree

Walk the examples with the human or cite an existing authoritative specification. Mark unresolved work as blocked; preserve the link from requirement to later verification.

Record these artifacts after doing the work:

  • The requirement has an attributable agreement or authoritative source. checklist check agree agreement --run "$RUN" --evidence "<artifact and observation>"
  • Implementation and test work can trace back to acceptance examples. checklist check agree handoff --run "$RUN" --evidence "<artifact and observation>"

When this stage is fulfilled: checklist advance agree --run "$RUN".

Branches

For a small change, a few concrete examples may be the entire artifact. If the human wants possibilities rather than a commitment, use an appropriate heuristic lens instead of manufacturing a specification.

If an input or permission is missing, name the blocker and leave the relevant item pending. A changed requirement may need a new run; an accepted checklist edit uses resume "$RUN" --refresh and requires fresh readings. To abandon work, use checklist reset --run "$RUN" --reason "<reason>"; history is retained.

Completion

An agreed task definition, or a clearly blocked draft naming the missing decision. A checklist confirmation records the agreement source; it does not grant consent.

After all required work and stage closures is recorded, use checklist done --run "$RUN". This archives the result without deleting its history; it does not authorise any external action.

Reference shelf

Open the technique relevant to the current step. These notes provide concrete methods and examples, not additional phases; this entrypoint defines the run's completion conditions.

  • Requirements that are easy to leave implicit
  • Separate the desired outcome from proposed implementation
  • Ask only what changes the work
  • Elicit examples and constraints from actual work
  • Write a task definition someone can verify
  • Keep agreement and later changes attributable

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.