agentleFS
gigio1023gigio1023/agent-skills/skills/minimal/terminology/SKILL.md

terminology

Skill2 starsChanged 11 days agoMITView on GitHub

Look up and apply a project's recorded terms, local names, and wording decisions, and settle which name the project uses when names compete. Use when working in a project with a terminology.md index, when the user asks what a term means or which name to use, when a new, contested, or misused term comes up, or when the user corrects or bans a wording: "what do we call this", "which name should we use", "이 용어 뭐라고 써", "용어 정리해줘". Settling a name surveys established projects, papers, and products and records the choice with sources. NOT for adding undisputed field terms, or for sentence style and claim strength in documents (copydesk).

5 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

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 6

    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:

    When the user asks about a term, answer with the adopted name, its meaning here, what other sources call it, and the reference.

    line 17
  2. Distinctive:

    When none holds, mention a suspected gap in one line and keep working with the plain description.

    line 31
  3. Distinctive:

    Start from what lookup found.

    line 45
  4. Distinctive:

    Write the row and its reference records as in records and layout.

    line 49
  5. Distinctive:

    Write records in the project being worked on.

    line 61
  6. Distinctive:

    After curation, check that every new or changed row cites a reference record that exists and that the root index reaches every topic file.

    line 77

“Common” is measured: kept by at least 6 in 10 of the 3 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

Terminology

A project glossary answers three questions per concept: which name the project uses, what it means here, and what others call it. It is a short lookup table, not a writing guide. Terms stay in the field's standard language, usually English; meanings are written in the reader's language.

Lookup is the ordinary path: read the glossary and apply it. It is cheap, writes nothing, and never starts a source survey. Curation is the second path and runs only under the conditions in "When to curate" below.

Lookup

Read root terminology.md and only the topic files the task touches. Use each adopted name with its recorded meaning in answers, documents, and code explanations. Keep confirmed local names for their recorded audience, and honor wording decisions within their scope and exceptions. Quotations, code identifiers, APIs, and data labels keep their original form.

When the user asks about a term, answer with the adopted name, its meaning here, what other sources call it, and the reference.

A field term with no row is usually undisputed: use it in its standard sense and move on. When the task needs a name the glossary does not settle, describe the thing plainly, and keep an unconfirmed local name as observed without asserting its meaning. A project without terminology.md has nothing to look up; curation creates one only when a curation condition holds.

When to curate

Curate only when the current task needs the name and one of these holds:

  • The glossary has no row for it and the name is in doubt: you see competing names, one name carrying different meanings, or no established name at all.
  • The project uses a name with a local meaning that differs from the field's.
  • A recorded entry looks wrong or stale.
  • A local name has no recorded confirmation.
  • The user corrects or bans a wording, or asks to add, check, change, or prune entries.

When none holds, mention a suspected gap in one line and keep working with the plain description. In a read-only task, curation returns the proposed row instead of writing it.

Inclusion test

A concept gets a row only when at least one of these holds:

  • Names compete, or one name carries different meanings across industry projects, papers, or competitor products.
  • A company or local meaning differs from the field's meaning.
  • The user made an explicit wording decision, such as banning a word.

Undisputed field terms stay out even when the project uses them constantly. Sentence patterns, "do not write X" advice, and project explanations are not rows. Claim strength and sentence style belong to copydesk.

Survey and decision

  1. Start from what lookup found. A correct entry needs no work. A wrong or stale one is updated in place and keeps its reference ID.
  2. Find how two or three recognizable sources in each relevant class name the concept, and whether their meaning matches: established projects, defining papers, and competitor products. Naming survey says which sources count and how to settle a split.
  3. Adopt one name in its established spelling. When no established name exists, describe the thing plainly and mark the row unresolved instead of coining one.
  4. Confirm local names with the user before recording them, as in local names.
  5. Write the row and its reference records as in records and layout.

Entry format

TermMeaningElsewhere calledReference
subagent definitionThe configured record that fixes a subagent's model, tools, and instructionsCodex: role; Claude Agent SDK: AgentDefinitionR002, R004

Meaning is one line. Elsewhere called lists source: name pairs, including names the project did not adopt. A one-line caveat goes under the table only where misuse actually recurs. Local names and wording decisions have their own tables.

Scope of changes

Write records in the project being worked on. After adopting or changing a name, update the documents the current task already edits and list other affected documents rather than sweeping the repository. Quotations, code identifiers, APIs, and data labels keep their original form.

Project instruction snippet

A skill description does not load this skill on every task; the project's instruction file does. Lookup never edits instruction files. During curation, or when the user asks to set up terminology for the project, add this snippet to the instruction file the project's harnesses read, such as AGENTS.md or CLAUDE.md, if none carries it yet. When that file is outside the task's write scope, return the snippet as a proposal.

## Terminology

Read `terminology.md` at task start and use its recorded names, local names, and wording decisions. When a needed term is missing, contested, or wrong, or a local name is unconfirmed, use the `terminology` skill to settle it.

Link the glossary rather than copying it into the instruction file.

Finish

After curation, check that every new or changed row cites a reference record that exists and that the root index reaches every topic file. Report adopted or changed names, rows left unresolved, and local names still waiting for the user.

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.