agentleFS
hoophqhoophq/hoop/.claude/skills/shape/SKILL.md

shape

Skill848 starsChanged 24 days agoMITView on GitHub

Turn a dictated idea or a Linear project into ready-to-run tickets - investigate the code, cut vertically, size by risk, write them into Linear. Use before /build-ticket, when the work is not a ticket yet.

  • Reads credentials

38 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use layout, explore and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • owshq-mecSkill

    layout

    Improve layout, spacing, and visual rhythm. Fixes monotonous grids, inconsistent spacing, and weak visual hierarchy. Use when the user mentions layout feeling off, spacing issues, visual hierarchy, crowded UI, alignment problems, or wanting better composition.

    Also in 8 projects that use this

  • genomewalkerSkill

    explore

    RLM-style recursive memory exploration - dynamically navigate the memory graph

    Also in 3 projects that use this

  • Microsoft AzureSkill

    build

    Build the TypeSpec Rust emitter from source. Run this after cloning the repo, after pulling changes, or when TypeScript source files have been modified.

    Also in 5 projects that use this

  • AnthropicSkill

    brief

    Generate contextual briefings for legal work — daily summary, topic research, or incident response. Use when starting your day and need a scan of legal-relevant items across email, calendar, and contracts, when researching a specific legal question across internal sources, or when a developing situation (data breach, litigation threat, regulatory inquiry) needs rapid context.

    Also in 3 projects that use this

  • jaydubya818Skill

    why

    Use when design rationale must be recovered from source history and available records.

    Also in 3 projects that use this

  • NVIDIASkill

    research

    Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.

    Also in 7 projects that use this

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 36 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write every Linear field in English, whatever the conversation language.

    line 11
  2. Distinctive:

    Do this before cutting anything.

    line 21

“Common” is measured: kept by at least 6 in 10 of the 36 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

Shape work into tickets

Input: a dictated idea, a rough note, or a Linear project ID. Output: tickets in Linear that an agent can run without asking anything.

Write every Linear field in English, whatever the conversation language.

1. Understand

Restate the goal in two lines and say who it is for. If a Linear project was given, read it and its existing issues first, so you extend rather than duplicate.

2. Investigate the code

Do this before cutting anything. Use subagents in parallel for breadth. Answer:

  • Which modules: gateway/, agent/, client/, common/, tunnel/, sidecar/, agentrs/, webapp_v2/. The legacy webapp/ is frozen.
  • Does it touch the gateway↔agent contract (common/proto/transport.proto, packet types, spec keys, payload formats)? That decides major.
  • Does it touch the sidecar config the control plane serves?
  • Migration needed? Read gateway/migrations/ for the next free number.
  • New env var? The helm chart and .env.sample change in the same PR.
  • New or superseded analytics event?
  • Does it need an ADR in hoophq/adr? Only for a structural change that spans modules, is expensive to reverse, changes the wire contract, or picks between real alternatives.

3. Cut vertically

Each ticket must be one complete flow, thin, testable end to end on its own.

GOOD  PR1: the whole flow for one connection type, behind a flag
      PR2: widen it to the rest
BAD   PR1: migration   PR2: model   PR3: API   PR4: UI

A horizontal cut leaves nothing testable until the last ticket. If a slice cannot stand alone, put it behind a feature flag rather than splitting it by layer.

4. Size by risk, not by lines

Risk = how much existing behaviour the change touches, times how hard it is to reverse.

  • Its own small ticket: auth, transport.proto and packet types, migrations, middleware order, anything a released agent reads.
  • May be large: new isolated code behind a flag, or a mechanical refactor with tests. 3000 lines of new package is safer than 50 lines in AuthMiddleware.

5. When the direction is unclear, write a spike first

A spike builds the whole thing on a throwaway branch to learn, and delivers evidence, not a PR. Once the human approves the direction, shape the prototype into vertical tickets, using it as the specification.

Code is cheap now. Build big to learn, merge small to ship.

6. Write the tickets

One ticket per slice, in this template, with blocked by between them:

## Type
bug | feature | spike

## Goal
Why, and for whom. 1-3 lines.

## Definition of done
- [ ] Verifiable item (QA checks each one)

## Scope / Out of scope

## Context
Files, past PRs, ADRs, customer threads.

## Constraints
- Feature flag: yes/no, default true/false
- Old agents must keep working: yes/no
- Migration / API change / analytics event: yes/no

## Evidence expected
Screenshot of X, VHS of CLI flow Y.

## Open questions
Empty, or the agent investigates first.

## Checkpoint
none | approve design before code

Rules for filling it:

  • Definition of done is the contract QA checks. Every item must be verifiable by someone who did not write the code. No "works well".
  • Feature flag: answer it here so the agent never asks. A feature that only adds a new code path is born Default: true; one that changes existing behaviour is born false.
  • Checkpoint: none for a bug. approve design before code for anything that picks between real alternatives.
  • Reference a ticket by ID, never by URL.
  • Fewest possible words. Simplified Technical English.

7. Hand back

Report, in the chat: the ticket IDs with one line each, the order and dependencies, and anything you could not decide. Then say the command to run them, one worktree per ticket:

claude --worktree <ticket-id>    # then /build-ticket <TICKET-ID>

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.