toolchain-upgrade
/toolchain-upgrade
Aktualizuje lintery, formatery i narzędzia typów, porównując konfigurację przed i po, by uniknąć cichego dryfu.
- Autor
- cravingmaker ↗
- Repozytorium
- skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 10.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 22
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 10.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Aktualizuje lintery, formatery i narzędzia typów, porównując konfigurację przed i po, by uniknąć cichego dryfu.
Oryginalny opis (EN)
Upgrade linters, formatters, type checkers, their plugins, and other dev tooling without silent drift, by comparing the effective configuration before and after, deciding every changed rule explicitly, and checking compatibility. Use when bumping dev dependencies or lint, format, or type tooling, or when Dependabot or Renovate opens such an update.
Jak działa
- Steps
- Rules
Podgląd SKILL.md
--- name: toolchain-upgrade description: Upgrade linters, formatters, type checkers, their plugins, and other dev tooling without silent drift, by comparing the effective configuration before and after, deciding every changed rule explicitly, and checking compatibility. Use when bumping dev dependencies or lint, format, or type tooling, or when Dependabot or Renovate opens such an update. license: MIT --- # Toolchain upgrade An upgrade is safe when every change in **effective behavior** is known and decided. A version bump that passes CI can still change what the tools enforce: a new rule turned on by a preset, a changed default, a removed rule, a different autofix. This skill makes those changes visible and deliberate. ## Steps 1. **Scope the upgrade.** List the packages to upgrade and their target versions. Respect the project's update policy: exact pins, minimum release age, and one upgrade per commit when the project asks for that. A target version that violates the policy waits. 2. **Capture the before state.** Record resolved versions (the lockfile, not the ranges) and the effective configuration. Use the project's own inventories first: a policy-diff script, rule-coverage or rule-option tests, a printed resolved config (`eslint --print-config <file>`, `tsc --showConfig`, `prettier --find-config-path` with the resolved options). Confirm which configuration that command consumes; a separate configuration file alone does not prove enforcement. 3. **Read the release notes** for every version between the current and the target, from the official changelog or release page. Note breaking changes, new and removed rules, changed defaults and options, preset changes, autofix changes, and runtime or peer requirements. 4. **Upgrade** the packages and refresh the lockfile with the project's package manager. 5. **Capture the after state** the same way as step 2 and compare. Account for every difference: - added, removed, deprecated, and renamed rules - changed options and defaults - preset membership changes - autofix behavior changes - runtime and peer-dependency compatibility, including overrides and resolutions that now pin something stale 6. **Decide each difference.** Accept it, configure it explicitly, or reject the upgrade. When the project's policy says new rules stay off until the policy covers them, set them off explicitly. Record the reason for each decision. 7. **Verify.** Run the project's full checks, focused tests for any behavior that changed, and consumer checks when the project is a package others install (pack it and import it in a clean consumer). 8. **Report** with [references/upgrade-report.md](references/upgrade-report.md). Done when every difference between the before and after states has a recorded decision, the full checks pass, and the report lists anything left unverified. Say the effective behavior is unchanged only when the comparison output is empty, and quote that output; a difference you configured back to its old value is still a decision to record. ## Rules - One concern per change: the upgrade and its configuration decisions land together; unrelated fixes land separately. - A failing check after an upgrade gets a decision, not a blanket disable. Disabling a rule is a recorded policy decision with a reason. - When release notes are missing or unclear for a version, say so in the report and rely on the before-and-after comparison. - Bind each check claim to what the check actually reads and enforces. Manifest/lockfile agreement does not prove peer compatibility or synchronized package versions; inspect the check and test the proposed version combination before predicting that it fails. Report compatibility not exercised by that check as unverified.
Pliki w paczce (22)
- SKILL.md 4 KB
- evals/examples.md 1 KB
- evals/results/2026-11-08-claude-claude-opus-5-5-activation-2-lean-2-toolchain.md 4 KB
- evals/results/2026-11-08-claude-claude-opus-5-5-activation-2.md 4 KB
- evals/results/2026-11-08-claude-claude-opus-5-5-near-miss-2.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-activation-1-final-evals.md 5 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-activation-1-round-2.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-activation-1.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-activation-2-final-evals.md 5 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-activation-2-round-2.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-near-miss-1-final-evals.md 5 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-near-miss-1-round-2.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-near-miss-1.md 4 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-near-miss-2-final-evals.md 5 KB
- evals/results/2026-11-08-codex-gpt-6.1-sol-near-miss-2-round-2.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-activation-1-claude.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-activation-1-round2-pre-claude.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-activation-2-claude.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-activation-2-round2-pre-claude.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-near-miss-1-claude.md 4 KB
- evals/results/2026-11-09-claude-claude-opus-5-5-near-miss-2-claude.md 4 KB
- references/upgrade-report.md 1 KB
Wersje
W bazie nie ma innych skilli o nazwie toolchain-upgrade.
Migawka bazy z 13.11.2026 · wersja paczki 0720cd045bf9 (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
- 100setup
Konfiguruje lintery, formatery, hooki gita i sprawdzanie typów, wykrywając język projektu: JS/TS, Python, Go lub Rust. - 100deps-upgrade
Aktualizuje zależności z wykrywaniem zmian łamiących i planem migracji, w tym podbicia wskazanych pakietów oraz przegląd PR-ów Dependabota i Renovate. - 95hk-configure
Konfiguruje haki git i kontrole projektu w pliku hk.pkl: dodaje lintery, formatery i własne kroki. - 100linter-setup
Konfiguruje lintery, formattery, kontrolę typów i analizę statyczną w bramkach pre-commit oraz CI. - 100linter-setup
Konfiguruje lintery, formattery, kontrolę typów i analizę statyczną w bramkach pre-commit oraz CI. - 90python-developer-tooling-build
Statycznie przegląda narzędzia deweloperskie i konfigurację builda w Pythonie: czy lintery, type-checkery i testy łapią realne defekty, pokrycie bramek CI, izolację tox i nox, układ projektu i pre-commit.