---
title: "Glossary"
canonical: glossary.html
---

<a id="glossary"></a>

# Glossary

- **answer** — what became of a feedback: the change, and the commit that made
  it.
- **archive** — to move a feedback to `feedback/archive/` once it has an
  answer.
- **catalog** — the component catalog in `knowledge/catalog/`.
- **checkout** — one TYPO3 core worktree below `.checkouts/`, this
  repository's own, created by `bin/cli checkouts:update`.
- **contract case** — a targeted scenario in `scenarios/contracts/`, which
  names one task shape.
- **decision** — what a change rested on, and what would show it wrong: one file
  in `decisions/`.
- **a feedback** — one report about this server, written by a session working
  somewhere else. Countable here: one report, one file, one subject.
- **the feedback channel** — `feedback/`, where feedback arrives and waits.
  Written by `typo3_feedback_record`, read by `bin/cli feedback`.
- **forward review** — an open scenario in `scenarios/forward/` that names no
  subsystem, skill or expected finding.
- **hint** — one statement in `knowledge/hints/`, reached by
  `bin/cli hints:probe`.
- **hold** — two senses, told apart by the preposition. A statement *holds on* a
  version or *for* a scope: it is true there. A test *holds* a rule and the rule
  is *held by* it: the test fails when the rule breaks.
- **installation** — the TYPO3 a caller works in. The server reads its own facts
  from it rather than bundles them.
- **judge** — to work out what should become of one open feedback, on evidence.
- **knowledge** — everything below `knowledge/`: what the tools answer from.
- **record** — the verb a feedback arrives by, and one of the six tool verbs.
- **requirement** — what must be true from now on, and what holds it there: one
  file in `requirements/`.
- **result** — what a tool hands back: the text and the same answer as data.
  `src/Result/` holds what several tools build one from — the shared schemas,
  the renderers, the unanswered case.
- **run** — one recorded execution of a forward review, in `scenarios/runs/`.
- **scenario** — one prompt and what has to come out of it, in `scenarios/`.
- **skill** — one canonical task workflow under `skills/`, installed into
  somebody else's project.
- **standalone checkout** — this repository as the Composer root package, where
  a session can record feedback.
- **todo** — one entry in the order of the work.
- **tool** — one `typo3_<subject>_<verb>` this server offers its callers: the
  MCP primitive of that name, one class in `src/Tool/`, listed in
  `TYPO3\DevCompanion\Tools`.
- **tool verb** — one of `lookup`, `guide`, `list`, `scope`,
  `describe`, `record`: what tells a caller the shape of the answer.
- **verdict** — how a recorded run came out: `covered`, `partial`, `gap`.