fluent-tui-knowledge
/fluent-tui-knowledge
Sprawdza polecenia i gotowe zestawy interfejsu tekstowego Ansys Fluent, podając źródło każdej informacji.
- Autor
- AstDen ↗
- Repozytorium
- kady-fluent-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 12.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 12.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Sprawdza polecenia i gotowe zestawy interfejsu tekstowego Ansys Fluent, podając źródło każdej informacji.
Oryginalny opis (EN)
Looks up Ansys Fluent TUI commands and pack recipes, reporting status, source and index path for every fact. Use when the user asks whether a command exists or how something is enabled in the TUI. Not for choosing models, writing or judging journals, or running Fluent.
Jak działa
- When to use / when not to
- Inputs
- Procedure
- Answer format
- Outputs
- Refusals
- Hand-off
Podgląd SKILL.md
--- name: fluent-tui-knowledge description: Looks up Ansys Fluent TUI commands and pack recipes, reporting status, source and index path for every fact. Use when the user asks whether a command exists or how something is enabled in the TUI. Not for choosing models, writing or judging journals, or running Fluent. license: MIT compatibility: Node ≥ 22.19 (bundled with Kady). No network. No Fluent required for this skill. metadata: version: "0.1.0" pack: fluent-skill-pack level: L1 --- # Fluent TUI knowledge — verified lookup This skill answers one question at a time: does the offline pack know this TUI fact, and with what evidence? You own the lookup and the honesty of the label on it — never a physics decision and never a journal. Two read-only CLIs search the vendored command index and the 72 recipes; every statement you make about Fluent comes from their JSON, not from memory. ## When to use / when not to - Use: "такая команда есть?", "как включить k-omega SST?", "какие параметры у pressure-inlet?", "есть рецепт для wall thermal?", "что делает этот путь?". - Do not use: a model / solver / BC choice → `fluent-model-selection`; journal text → `fluent-journal-authoring`; "журнал правильный?" → `fluent-journal-validation`; running or monitoring Fluent → `fluent-controlled-run` (this skill never runs anything). ## Inputs - the user question, in any language; - the pack home `<home>`: `<project>/sandbox/.kady/fluent`, or whatever `KADY_FLUENT_HOME` points at. The CLIs resolve it themselves — never assemble or guess a path; - the two CLIs, invoked exactly like this: ```bash node <home>/bin/kb-search.mjs --query "k-omega SST" --scope all --limit 10 node <home>/bin/kb-search.mjs --query "pressure inlet" --category boundary node <home>/bin/kb-get.mjs --path recipes/models.viscous.kw-sst node <home>/bin/kb-get.mjs --path /define/models/viscous --limit 40 node <home>/bin/kb-get.mjs --path docs/CASE_SPEC.md --offset 1 --limit 60 node <home>/bin/kb-get.mjs --path recipes/ # list every recipe id with its status node <home>/bin/kb-get.mjs --path docs/ # list the bundled documents ``` Search flags: `--scope all|commands|recipes`, `--limit 1..100`, `--category <substring>`. `kb-get` pages documents, command rows and listings with `--offset/--limit` and sets `truncated` plus `nextOffset`. Both print exactly one JSON object on stdout and diagnostics on stderr; exit codes are `0` ok (an empty hit is still ok), `2` usage, `3` unknown path/recipe or missing pack data, `4` refused path (outside the pack). ## Procedure 1. Turn the question into 2–4 English TUI terms (the index and the recipes are English); keep the user's own phrasing as a second query when it looks like a command path. 2. Run `node <home>/bin/kb-search.mjs --query "<terms>" --scope all --limit 10` and read `counts` plus `results`. Ranking puts exact/prefix path and recipe-id matches first, then token coverage, so row one is a candidate, not a proof. 3. Confirm the candidate with `kb-get.mjs`: a recipe through `--path recipes/<id>` (its `status`, `source`, `journal[]`, `params[]`), a command path through `--path <that path>` (the row itself plus every row beneath it). Never answer from a search snippet alone. 4. Page long documents instead of truncating them: `kb-get.mjs --path docs/<file> --offset N --limit M`, then follow `nextOffset` while `truncated` is true. 5. Label every fact with the table below and answer in Russian, one fact per sentence. 6. If the search is empty, retry once with terms taken from the closest recipe titles, then report the miss with its `hint` and name the live check a Fluent session would need. ## Answer format Every factual sentence carries three fields, copied from the JSON, never reconstructed: | Fact | `status` | `source` | `path` | |---|---|---|---| | the TUI accepts this path | `index` | `index/commands.csv`: `category`, `kinds`, `passes` | `path` verbatim (a trailing `?` asks yes/no, with `default`) | | a recipe covers this setup step | recipe `status`: `verified` \| `derived` \| `unverified` | recipe `source` (`KB/…:line`) | `recipes/<id>` | | reference material | — | the document itself | `docs/<file>` | - `index` means "present in the vendored command index", not live verification: about 61 % of indexed paths were never exercised in a live session, so say "в индексе есть, вживую не проверено". - `verified` / `derived` / `unverified` belong to the recipe. Never promote one, never average them, never call a command verified. - Quote `question`, `default` and `blocked` instead of paraphrasing a row; `blocked: yes` means some menus refuse the command — mention that when it is set. - When another skill could also answer, hand the fact over with its label rather than re-deriving it. ## Outputs A chat answer only. Nothing is written to disk: no cache, no notes, no journal, no task state. ## Refusals - **No invented syntax.** If both CLIs have no row, the pack does not know the command: say so, quote the empty-result `hint`, and name what a live session must verify. Never answer from memory. - **No silent promotion.** `index` never becomes `verified`; `derived` and `unverified` are shown as such in every sentence. - **No live claims.** You cannot see a running Fluent, so phrase it as "в индексе", never "работает". - **No writes.** Both CLIs are read-only; the pack forbids editing `lib/` and `data/`. - **No physics choice** (turbulence model, BC type, solver settings for a given Re/Ma) → route to `fluent-model-selection`; **no journal text** → `fluent-journal-authoring`. - **No network, no installs, no Fluent.** If the pack looks broken (`ok:false`, missing home), report the error text verbatim and stop. ## Hand-off Pass the concrete record, not a summary: `path`, `status`, `source` and the fields the next skill needs. `fluent-model-selection` receives the candidate recipe ids with their statuses; `fluent-journal-authoring` receives recipes and command paths with their `default` values; `fluent-journal-validation` receives the paths it must judge against. A fact the pack lacks is a `fluent-kb-maintenance` job (phase 3), never a licence to guess.
Wersje
W bazie nie ma innych skilli o nazwie fluent-tui-knowledge.
Migawka bazy z 13.11.2026 · wersja paczki 78c5636677ac (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 90fluent-case-builder
Kieruje zadania Ansys Fluent do właściwego wyspecjalizowanego skilla i zapisuje na dysku, na jakim etapie jest przypadek, także po wznowieniu rozmowy. - 95ansys-fluent-mcp
Steruje sesją solvera ANSYS Fluent przez serwer MCP: wczytuje przypadek i siatkę, zmienia ustawienia, uruchamia iteracje i odczytuje wyniki. - 90fluent-journal-validation
Ocenia dziennik Fluent offline względem indeksu poleceń, sprawdza kolejność i zakończenia wierszy i wydaje werdykt: czysty, ostrzeżenie albo zablokowany. - 85fluent-cfd
Prowadzi symulacje CFD w Ansys Fluent i PyFluent: konfigurację siatki, warunków brzegowych i turbulencji, zbieżność oraz walidację wyników. - 90run-fluent-autoclave
Prowadzi, diagnozuje i waliduje symulacje CFD w ANSYS Fluent dla autoklawów z konwekcją wymuszoną, geometrią STEP i kontrolą bilansu masy. - 90CFD Fluent Pipeline
Prowadzi audytowalną symulację CFD w Ansys Fluent i PyFluent: siatkę, warunki brzegowe, model turbulencji, zbieżność i walidację wyników.