菜单
Guided workflow for adding a new linter to MegaLinter. Use when a contributor needs to add support for a new linting tool.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/oxsecurity/megalinter --skill add-linter命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | add-linter |
| description | Guided workflow for adding a new linter to MegaLinter. Use when a contributor needs to add support for a new linting tool. |
| allowed-tools | Read Grep Glob Bash Edit Write WebSearch WebFetch |
| argument-hint | [linter-name] |
| model | sonnet |
| metadata | {"internal":true} |
Guide me through adding the linter $ARGUMENTS to MegaLinter. If no linter name was provided, ask me for:
Search the internet to gather all available information about the linter:
is_formatter: true) or a linter.eslintignore)Check if a descriptor exists in megalinter/descriptors/ for this language. If not, create a new <lang>.megalinter-descriptor.yml.
If the linter covers several descriptors (like eslint, prettier or biome), define it once in megalinter/descriptors/shared/<linter_name>.megalinter-linter.yml (a complete standalone linter definition) and add a thin extends: <linter_name> entry in each descriptor with only the per-descriptor overrides (test_folder, examples, file_extensions…) — see .claude/rules/descriptors.md → "Shared Linter Definitions".
Add the linter entry with as many properties as possible. Even though the JSON schema only requires linter_name, linter_url, and examples, aim for maximum completeness. Fill in ALL of these when applicable:
Identity (required):
linter_name — CLI executable name, lowercaselinter_url — tool websiteexamples — CLI usage (with and without config file)Documentation (strongly recommended):
linter_text — rich markdown description: features, what it checks, when to use itlinter_repo — GitHub repository URLlinter_rules_url — URL listing all ruleslinter_rules_configuration_url — how to configurelinter_rules_inline_disable_url — how to suppress rules inlinelinter_rules_ignore_config_url — how to ignore fileslinter_spdx_license — SPDX license IDlinter_speed — 1 (very slow) to 5 (very fast)linter_image_url, linter_icon_png_url, linter_banner_image_url — logos/bannersCLI configuration (always fill):
cli_lint_mode — default mode: file, list_of_files, or project
supported_cli_lint_modes — every mode the linter can run in (subset of file / list_of_files / project; defaults to ["file"], must include cli_lint_mode). Success/failure tests run once per declared mode, so only list modes the tool actually supports
cli_executable — if different from linter_name
config_file_name — default config file (e.g., .pylintrc)
cli_config_arg_name — config argument (e.g., --config)
cli_lint_extra_args — default extra arguments
cli_lint_fix_arg_name — fix argument (e.g., --fix)
cli_lint_fix_remove_args — args to remove in fix mode
cli_version_arg_name — if not --version
cli_help_arg_name — if not --help
ignore_file_name, cli_lint_ignore_arg_name — ignore file support
Excluded directories forwarding — when project is a supported lint mode, MegaLinter must forward EXCLUDED_DIRECTORIES to the tool or it will scan node_modules/build caches raw. Pick exactly ONE mechanism (see .claude/rules/descriptors.md → "Project Lint Mode: Forwarding Excluded Directories" for full semantics and known traps):
cli_lint_mode_project_exclude_arg_name (+ _arg_value {{DIR}}/{{WORKSPACE}} template, _separator if a repeated flag overrides, _seed_values if the flag replaces the tool's built-in defaults, _config_key if it replaces a list in the tool's config file)cli_lint_mode_project_exclude_ignore_file_arg_name (+ _seed_files, _pass_existing)manage_excluded_directories_config() override in the linter classlinter_textResearch the official docs first: exact flag, value syntax (path/glob/regex, anchoring), repeatability, and whether it replaces config/built-in defaults — a wrong choice silently drops exclusions or clobbers user configuration.
Poison fixture — after declaring forwarding, add a deliberately failing file in .automation/test/<test_folder>/good/.wireit/ so test_success_project_lint_mode guards the forwarding against regressions (only if all project-capable linters sharing the folder have forwarding)
Error parsing (important for accurate counts):
cli_lint_errors_count — regex_count, regex_number, regex_sum, total_lines, or sarifcli_lint_errors_regex — regex matching error lines in outputSARIF support (if available):
can_output_sarif: truecli_sarif_args — with {{SARIF_OUTPUT_FILE}} placeholderBehavior flags:
is_formatter: true — if it's a formatteractivation_rules — if it depends on env vars (e.g., style preference)active_only_if_file_found — only activate if certain config files existInstall (required):
install:
dockerfile:
- |-
# renovate: datasource=pypi depName=tool-name
ARG PIP_TOOL_VERSION=1.2.3
pip:
- tool-name==${PIP_TOOL_VERSION}
supported_platforms with install_override for ARM if neededIDE section (always fill):
ide.vscode, ide.idea, ide.eclipse, ide.sublime, ide.emacs, ide.atom, ide.visual_studio[{name: "Extension Name", url: "marketplace-url"}]Testing:
test_folder — if different from lowercase descriptor_idtest_variables — env vars for testsLook at existing well-populated descriptors like megalinter/descriptors/python.megalinter-descriptor.yml for reference.
Create two test files in .automation/test/<test_folder>/:
cli_lint_errors_regexCreate a minimal class in megalinter/linters/ extending megalinter.Linter. Only override what the YAML descriptor can't express.
Run make megalinter-build to auto-generate Dockerfiles, test classes, schemas.
Do NOT run make megalinter-build-with-doc — documentation is handled by auto-update workflows and generating it in PRs causes merge conflicts.
LINTER="<descriptor_id_lowercase>_<linter_name>"
docker buildx build --platform linux/amd64 --file linters/$LINTER/Dockerfile --tag $LINTER .
docker run --rm --env TEST_CASE_RUN=true --env OUTPUT_DETAIL=detailed \
--env TEST_KEYWORDS="${LINTER}_test" --env MEGALINTER_VOLUME_ROOT="." \
--volume "$(pwd):/tmp/lint" $LINTER
CHANGELOG.md (repo root, beta section):
- Add [linter-name](linter_url) linter for <language> — <what it detects, one sentence>
user/add-<linter-name>quick build + TEST_KEYWORDS=<linter>_test in commit message body during dev