misata
/misata
Generuje relacyjne dane testowe i demonstracyjne ze zadeklarowanego schematu w Misata, z kluczami obcymi, regułami UNIQUE i CHECK oraz krzywymi przychodów, a następnie atestuje pliki.
- Autor
- rasinmuhammed ↗
- Repozytorium
- misata ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 26.10.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 31
- Skrypty
- 1
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 1
Brak sygnałów. Repozytorium zmienione 26.10.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
- 1 skrypt w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.
Skrypty: docs_seo_hook.py
Co robi i kiedy
Generuje relacyjne dane testowe i demonstracyjne ze zadeklarowanego schematu w Misata, z kluczami obcymi, regułami UNIQUE i CHECK oraz krzywymi przychodów, a następnie atestuje pliki.
Oryginalny opis (EN)
Generate relational test, demo or evaluation data from a declared schema with Misata, where every declaration (foreign keys, unique and CHECK rules, monthly revenue curves, rates, group shares, ledger identities) holds exactly or is refused before generation, then attest the files. Use when a task needs a seeded database, fixtures that load without constraint errors, a demo or BI dataset with believable shapes, or a text-to-SQL eval with known answers, and there is no real data to use.
Jak działa
- Install
- The loop
- Reading results
- MCP tools
- Do not claim
Podgląd SKILL.md
--- name: misata description: Generate relational test, demo or evaluation data from a declared schema with Misata, where every declaration (foreign keys, unique and CHECK rules, monthly revenue curves, rates, group shares, ledger identities) holds exactly or is refused before generation, then attest the files. Use when a task needs a seeded database, fixtures that load without constraint errors, a demo or BI dataset with believable shapes, or a text-to-SQL eval with known answers, and there is no real data to use. --- # Misata Misata turns a schema into data that holds what the schema declares. A declaration it cannot honour is refused before any row exists, with the arithmetic and a remedy. Full guidance for agents is in [AGENTS.md](https://github.com/rasinmuhammed/misata/blob/main/AGENTS.md). ## Install ```bash pip install misata # or run without installing: uvx misata --help pip install "misata[mcp]" # the MCP server, misata-mcp ``` ## The loop 1. **Write the schema.** From DDL: `misata.from_ddl(ddl)`. From a live database: `misata init --db URL`. From scratch: `misata init` writes a `misata.yaml` to edit. Always set a seed. 2. **Plan.** `misata explain misata.yaml --json` returns planned rows per table, which declaration writes which column, and every conflict. Exit 3 means the declarations cannot all hold: read `conflicts`, apply each `remedy` in the schema, run it again. 3. **Generate.** `misata generate -c misata.yaml -o out --attest --json`. 4. **Report what the attestation says.** `data.attestation.result` is `PASSED` only when every declared invariant holds in the files on disk. `out/verify.py` lets anyone recheck it with only `duckdb`. When the numbers that matter come out of a pipeline (net revenue in a dbt model, a SQL metric) rather than off a column, declare an exact outcome curve on the column that drives them and run `misata solve -c solve.yaml -o out --json`. Exit 0 means every period is within tolerance; otherwise `data.solve.reason` and `message` say why it stopped and what to change. ## Reading results Every command takes `--json` and prints one envelope on stdout: `ok`, `code`, `exit_code`, `message`, `conflicts`, `data`, `warnings`. Exit codes: 0 ok, 1 a check failed, 2 bad arguments or unreadable input, 3 the declarations cannot all hold and nothing was written. A `ConformanceWarning` in `warnings` means a declaration bent; its message says which and why. Fix findings in the schema, never by editing rows. ## MCP tools `explain_schema` before generating, `generate_from_schema` to generate, `solve_outcome` to hit targets computed by SQL over the data, `attest_dataset` to prove the output, `check_realism` for the statistical tells of generated data, `seed_database` to fill a real database (it plans first; never pass `truncate` without the user's consent). ## Do not claim - that `mimic()` output is private or anonymous: it is fitted to real rows; - that the data is indistinguishable from real data; - byte-identical output across Misata versions (within one version, yes).
Pliki w paczce (31)
- .env.example 1 KB
- .gitignore 4 KB
- AGENTS.md 7 KB
- CHANGELOG.md 267 KB
- CITATION.cff 1 KB
- CODE_OF_CONDUCT.md 1 KB
- CONTRIBUTING.md 4 KB
- DATA-PROVENANCE.md 5 KB
- Dockerfile 2 KB
- FEATURES.md 6 KB
- LANGUAGE.md 25 KB
- LICENSE 1 KB
- LIMITATIONS.md 13 KB
- MISATA_GLOSSARY.md 5 KB
- PYPI.md 24 KB
- QUICKSTART.md 6 KB
- README.md 25 KB
- SECURITY.md 5 KB
- SKILL.md 3 KB
- STABILITY.md 4 KB
- docs_seo_hook.py 1 KB
- funding.json 5 KB
- glama.json 1 KB
- llms-full.txt 13 KB
- llms.txt 8 KB
- mkdocs.yml 7 KB
- pyproject.toml 7 KB
- requirements-specbench.txt 1 KB
- server.json 1 KB
- smithery.yaml 6 KB
- uv.lock 785 KB
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · rasinmuhammed/misata | 100 | MIT |
| rasinmuhammed/misata | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki 125329f0439e (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
- 100testing-dbt-models
Dodaje testy schematu i walidację jakości danych do modeli dbt w plikach schema.yml: unique, not_null, accepted_values, klucze i relacje, zgodnie z istniejącym stylem projektu. - 100seed
Generuje deterministyczne dane przykładowe z schematu bazy z kluczami obcymi w kolejności zależności, bez zapisu do bazy.