agentleFS
probabl-aiprobabl-ai/skore-skills/skills/shape-user-idea/SKILL.md

shape-user-idea

Skill144 starsChanged 17 days agoBSD-3-ClauseView on GitHub

Shape a user idea or a named artifact into one idea file and one Ideas row after they confirm. A topic with no artifact belongs to search-ml-literature. Do not write History, Backlog, or a design note. Do not pick a backlog row.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 2 are distinctive, and 7 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 7

    Very specific

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

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

    Run python -m skore_skills status.

    line 28
  2. Distinctive:

    If the entry is not already a concrete idea, question, or artifact, AskUserQuestion: free-text or artifact.

    line 31
  3. Very specific:

    Leave an unknown metric as n/a.

    line 35
  4. Very specific:

    If the idea changes the project goal in journal/JOURNAL.md Status, ask before writing the file.

    line 52
  5. Very specific:

    Do not write History, Backlog, Status, or a design note.

    line 88
  6. Very specific:

    Do not append a B<N> row. The backlog triages the file.

    line 90
  7. Distinctive:

    Do not invent source content from a title.

    line 91
  8. Very specific:

    Do not WebSearch for a topic with no artifact.

    line 92
  9. Very specific:

    If the user already asked to add a named library, put it in What changes and load add-python-package when that skill is installed.

    line 95

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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 User Idea

Comes after review, beside the backlog. The user already has an idea, a question, or an artifact. This skill writes journal/ideas/<slug>.md and one Ideas row only after they confirm. The backlog triages that file later.

Human-facing prose

Details: setup-workspace references/human_facing_prose.md. Questions and the confirmation paragraph describe this idea in data-science terms — not skill ids or the wrapper CLI.

Procedure

  1. Run python -m skore_skills status. Read journal/JOURNAL.md Status and History, and data_analysis/data_analysis.md when it exists.
  2. If the entry is not already a concrete idea, question, or artifact, AskUserQuestion: free-text or artifact. A URL, issue, path, or stated idea skips that menu.
  3. A question about this project is answered from those files. Leave an unknown metric as n/a. Write no idea file and return to the caller.
  4. A topic with no artifact ("what do people do for …") loads search-ml-literature only if status.skills.search-ml-literature is true. Otherwise one-line skip. Do not WebSearch for a paper from here.
  5. An artifact is read, not reconstructed. Paper or doc URL: WebFetch the page. GitHub issue: gh auth status first; if unauthenticated, ask the user to run gh auth login or paste the body. Then gh issue view. Resolution: explicit URL, then org/repo#N, then bare #N only in the current gh repo — otherwise ask. Spec or notes file: read that file only.
  6. Walk three questions. Do not fabricate a missing answer.
    • What are we trying to learn?
    • Why now? Quote the user or the artifact.
    • What changes versus the last experiment, in prose?
  7. If the idea changes the project goal in journal/JOURNAL.md Status, ask before writing the file. A library the source names and the user has not agreed to add stays in Open gaps. A domain claim this dataset does not establish stays in Open gaps.
  8. Restate the idea in one short paragraph, then one single-choice AskUserQuestion: Yes / No. Do not also ask for a typed yes. Put that question in the message before any idea file. "Maybe" is not Yes. No or Stop writes nothing. A Yes from the tool, or an explicit yes the user already gave, authorizes the write only after that question is in the message.
  9. On Yes, write journal/ideas/<slug>.md and upsert its Ideas row. Do not create a design note. Do not edit History, Backlog, or Status. Write journal/ideas/<slug>.md:
# <slug>
- Experiment: <last History stem, or n/a>
- Source: user
- Triage: open
- Question:
- Why now:
- What changes:
- Open gaps:

Why now carries the quote, URL, or path. No acceptance criteria. Then upsert one ## Ideas row: Question as plain text, Status open, Experiment <last History stem, or n/a>, Source user. If that table is missing, insert it between History and Backlog. Do not rewrite the other sections. Return the path to the caller (manage-ml-backlog or triage-ml-task).

Stop conditions

  • Do not write History, Backlog, Status, or a design note. The Ideas table is the only JOURNAL.md edit.
  • Do not append a B<N> row. The backlog triages the file.
  • Do not invent source content from a title.
  • Do not WebSearch for a topic with no artifact.
  • The entry question and the Yes / No confirmation are required gates.
  • If the user already asked to add a named library, put it in What changes and load add-python-package when that skill is installed. Do not pixi add from here.

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.