Legwork

Crop an image to the given box (left, top, right, bottom).

Run a csvkit subcommand with optional arguments and optional stdin. - command: one of csvkit's CLI tools (e.g., "csvcut", "csvstat", "in2csv"). - args: list of additional CLI arguments, e.g., ["-c", "name,age"] or ["--help"]. - input_data: if provided, will be piped to the command's stdin (useful for commands that accept stdin).

Differentiate an expression with respect to a variable. Args: expr: Expression to differentiate, e.g., "sin(x)*x**2". var: Variable name to differentiate with respect to, e.g., "x". order: Order of the derivative (default 1). Returns: The derivative as a string.

Execute a SQL query using DuckDB. - sql: The SQL query to run. Can reference local files directly (e.g., SELECT * FROM 'data.csv';). - database_path: Optional path to a DuckDB database file. If not provided, uses an in-memory database. Returns: { "columns": [col_name, ...], "rows": [ {col_name: value, ...}, ... ] }

Evaluate an expression optionally with substitutions. Args: expr: Expression to evaluate numerically or symbolically. subs: Optional dict mapping variable names to values (strings parsable by SymPy), e.g., {"x": "pi/2", "y": "2"}. Returns: The evaluated expression or number as a string.

Execute a non-SELECT SQL statement (e.g., CREATE TABLE, INSERT, UPDATE, DELETE) against a SQLite database. Args: db_path: Path to a SQLite database file (it will be created if it does not exist). sql: The SQL statement to execute. params: Optional dictionary of named parameters for the SQL statement. Returns: The number of rows modified if available, otherwise 0.

Extract main text (and optionally metadata) from a local HTML file. Args: file_path: Path to a local HTML file (read-only). output_format: One of "txt", "json", "xml", "html", "markdown". Defaults to "txt". with_metadata: If True and output_format="json", include metadata fields when available. Returns: - If output_format="json": a dict with extracted content (and metadata when available). - Otherwise: a string with the extracted content in the chosen format. Raises: FileNotFoundError if the file does not exist. RuntimeError if no content could be extracted.

Extract main text (and optionally metadata) from an HTML string. Args: html: A string containing the full HTML document to parse. output_format: One of "txt", "json", "xml", "html", "markdown". Defaults to "txt". with_metadata: If True and output_format="json", include metadata fields when available. Returns: - If output_format="json": a dict with extracted content (and metadata when available). - Otherwise: a string with the extracted content in the chosen format. Raises: RuntimeError if no content could be extracted.

Extract tables from a PDF using pdfplumber's table extraction. Arguments: - path: Path to a local PDF. - page: If provided, extract only this 1-indexed page; otherwise all pages. - table_settings_json: Optional JSON for table settings (see README "Table-extraction settings"). - laparams_json: Optional JSON of pdfminer.six LAParams to pass to pdfplumber.open(). - password, unicode_norm, strict_metadata: Passed to pdfplumber.open(). Returns: { "pages": [ { "page_number": int, "tables": [ [ [cell_str_or_null, ...], ... ] # rows ] } ] }

Extract text from a PDF (all pages or a single 1-indexed page). Arguments: - path: Path to a local PDF. - page: If provided, extract only this 1-indexed page; otherwise all pages. - layout: If True, attempt to preserve layout (experimental in pdfplumber). - x_tolerance, y_tolerance, x_tolerance_ratio: Passed to Page.extract_text(...). - laparams_json: Optional JSON of pdfminer.six LAParams to pass to pdfplumber.open(). - password, unicode_norm, strict_metadata: Passed to pdfplumber.open(). Returns: { "pages": [ { "page_number": int, "text": str|null } ] }

Convert an HTML file to Markdown and return the Markdown text.

Generate fake data using Faker by provider method name. - fake_name: provider method to call (e.g., "name", "address", "profile"). - locales: optional locale string or list of locales (e.g., "en_US" or ["it_IT", "en_US"]). - repeat: number of values to generate. - args: positional arguments for the provider method. - kwargs: keyword arguments for the provider method. - use_weighting: whether to use weighted real-world frequencies (default True). - seed: optional seed to make results deterministic per instance. - unique: ensure uniqueness for generated values in this call. - unique_locale: if …

Generate a QR code PNG image. - data: text to encode - out_path: file path to save (defaults to ./outputs/qr_<timestamp>.png) - version: 1..40 (size). If None, fit automatically. - error_correction: L (7%), M (15%, default), Q (25%), H (30%) - box_size: pixels per QR module - border: modules thick border (min 4) - fill_color, back_color: hex string/name or [R,G,B]

Generate a QR code SVG image. - data: text to encode - out_path: file path to save (defaults to ./outputs/qr_<timestamp>.svg) - factory: 'path' (default), 'basic', 'fragment', 'path-fill', or 'fill' - attrib: optional SVG root attributes dict (e.g., {'class': 'my-css'}) - version, error_correction, box_size, border: QR sizing options

Return PDF document metadata as a simple dict of string keys/values. Keys are normalized without the leading '/' when present.

Return low-level objects for a single page: chars, lines, rects, curves, images, annots, hyperlinks. Arguments: - path: Path to a local PDF. - page: 1-indexed page number. - types: Optional subset of object types to include. Allowed values (singular): "char", "line", "rect", "curve", "image", "annot", "hyperlink". If omitted, includes all. - precision: If provided, round floating-point numbers to this many decimal places. - laparams_json: Optional JSON of pdfminer.six LAParams to pass to pdfplumber.open(). - password, unicode_norm, strict_metadata: Passed to pdfplumber.open(). Retur…

Return basic metadata and page count for a PDF. Arguments: - path: Path to a local PDF file. - laparams_json: Optional JSON string for pdfminer.six layout analysis parameters. - password: Optional password for encrypted PDFs. - unicode_norm: Optional Unicode normalization form ("NFC","NFD","NFKC","NFKD"). - strict_metadata: If True, raise if invalid metadata is encountered. Returns: { "metadata": {..}, "page_count": N }

Get basic info about a PPTX: slide count and slide titles (if any). Args: pptx_path: Path to the PPTX file. Returns: Dict with keys: - slide_count: number of slides - titles: list of title texts (empty string if none on a slide)

Fetch the transcript for a YouTube video. Args: video_id: The YouTube video ID (not a full URL). languages: Optional list of language codes in descending priority (e.g., ["de", "en"]). preserve_formatting: If True, keep basic HTML formatting (e.g., <i>, <b>). Returns: A dict containing transcript metadata and a list of snippets with text, start, duration. Raises: RuntimeError: If retrieval fails.

Return the installed sqlfluff version string.

Get the gitleaks version string.

Convert an image to grayscale.

Compute the Gulpease index for Italian text.

Create a histogram and return the saved PNG image.

Convert a local HTML file to a PDF file. Args: html_file_path: Path to a local .html/.xhtml file. output_pdf_path: Optional path for the output PDF. If not provided, a file will be created under ./outputs/. base_url: Optional base URL or directory for resolving relative links (defaults to the input file's directory). Returns: A dict with: - pdf_path: absolute path to the generated PDF - size_bytes: file size in bytes - errors: number of conversion errors reported by xhtml2pdf (0 indicates none)

Convert an HTML string to Markdown using python-markdownify. All options mirror the library's documented options.

Convert an HTML string to a PDF file. Args: html: The HTML content to convert. output_pdf_path: Optional path for the output PDF. If not provided, a file will be created under ./outputs/. Returns: A dict with: - pdf_path: absolute path to the generated PDF - size_bytes: file size in bytes - errors: number of conversion errors reported by xhtml2pdf (0 indicates none)

Inspect an image and return basic information.

Read basic image metadata (format, width, height, etc.) using sharp. Returns a JSON-serializable dict of common fields.

Insert multiple rows into a table using sqlite-utils' insert_all, creating the table if needed. Args: db_path: Path to the SQLite database file (created if it does not exist). table: Table name to insert into (created if it does not exist). rows: List of dictionaries representing rows to insert. pk: Optional primary key column name to set on table creation. replace: If True, use OR REPLACE behavior for conflicts. upsert: If True, update existing rows when primary key conflicts occur. ignore: If True, use OR IGNORE behavior for conflicts. Returns: A summary dic…

Integrate an expression with respect to a variable, optionally with limits. Args: expr: Expression to integrate, e.g., "sin(x)". var: Variable name to integrate with respect to, e.g., "x". lower: Optional lower limit as a string (e.g., "0" or "pi/2"). upper: Optional upper limit as a string. Returns: The integral as a string.

Return the installed jsonschema library version string.

Lint a SQL string using sqlfluff and return the CLI output as text. Args: sql: The SQL code to lint. dialect: SQL dialect to use (e.g., 'ansi', 'postgres', 'snowflake', etc.).

List available transcripts for a YouTube video and their metadata. Args: video_id: The YouTube video ID (not a full URL). Returns: A list of dicts with fields: video_id, language, language_code, is_generated, is_translatable, translation_languages.

List available exporter short names provided by nbconvert.

List available fake provider method names for the given locales and optional custom providers.

List installed MarkItDown plugins (equivalent to `markitdown --list-plugins`).

List tables in the database, optionally including row counts. Args: db_path: Path to the SQLite database file. counts: If True, return list of dicts with table and count; otherwise return list of table names. Returns: A list of table names (strings) or a list of dicts with 'table' and 'count'.

Return the installed Matplotlib version.

Merge multiple PDFs into a single PDF. - pdf_paths: list of input PDF file paths (in order). - output_path: path for merged PDF. Defaults to ./merged.pdf. Returns the absolute path to the merged PDF.

The speech model this tool uses, and the faster-whisper version.

Return the installed MoviePy version.

Run OCR on a local image using RapidOCR (ONNX Runtime). - image_path: Path to a local image file (read-only). - visualize: If True, saves a visualization image with detected boxes and recognized text. Returns the recognized text, each line with its confidence and box, and optional elapsed time and visualization path.

Compute the OSMAN readability score (Arabic).

Draw simple text onto an image at (x, y) using the default bitmap font.

Parse an iCalendar (.ics) file from disk and return a summary of calendar-level properties and VEVENTs.

Parse iCalendar (.ics) content and return a summary of calendar-level properties and VEVENTs. Returns dict: { "calendar": {...}, "events": [ {summary, dtstart, dtend, ...}, ... ] }

Create a line plot and return the saved PNG image.

Check files for Prettier formatting issues without writing changes. Returns Prettier's combined output (stdout/stderr).

Format a code string with Prettier and return the formatted code. Parameters: - code: the source code to format - parser: optional Prettier parser (e.g., 'babel', 'typescript', 'json', 'markdown', 'html', 'css', 'yaml') - filepath_hint: optional pseudo filename to help Prettier infer the parser (e.g., 'file.ts', 'notes.md') - config_path: path to a Prettier config file to use - print_width, tab_width: numeric formatting options - use_tabs, single_quote, semi: boolean formatting options - trailing_comma: string option (e.g., 'none', 'es5', 'all')

Get the installed Prettier CLI version string.

Format and write changes to the given files in-place. Returns Prettier's output summary.

Generate ASCII art for a QR code and return it as text. - data: text to encode - border: modules thick border (min 0) - invert: swap foreground/background

Execute a read-only SQL query against a SQLite database using sqlite-utils and return rows as a list of dicts. Args: db_path: Path to a SQLite database file (it will be created if it does not exist). sql: The SQL query to execute. params: Optional dictionary of named parameters for the SQL query. Returns: A list of dictionaries, one per row.

Read all paragraph texts from a .docx file. Args: path: Path to an existing .docx file. Returns: {"paragraphs": ["para1", "para2", ...]}

Remove the background from an image file and save a PNG with transparency. - input_path: Path to the input image file. - output_path: Optional output path. If omitted, writes to ./outputs/<stem>.out.png. - model: Model name (e.g., 'u2net', 'u2netp', 'isnet-general-use', 'birefnet-general', 'bria-rmbg', etc.). Default 'u2net' for fast, small offline use. - alpha_matting: Enable alpha matting refinement (slower). - decontaminate: Enable edge color decontamination. - only_mask: Return only the mask (grayscale). - post_process_mask: Post-process mask to binary. - bgcolor: Background replacement c…

Resize and/or convert an image using sharp. - input_path: path to an existing image file. - output_path: where to write the result (extension may imply format). - width/height: target dimensions (either or both). - format: optional explicit output format (jpeg/jpg/png/webp/gif/avif). - fit: resizing fit mode (default inside). - auto_orient: rotate based on EXIF. - quality: optional quality for lossy formats (jpeg/webp/avif). Returns an Image that points to the written file.

Rotate an image by angle degrees. Use expand=True to grow the canvas to fit.

Run 'ruff check' to lint Python files. - paths: files or directories to lint; defaults to current directory if not provided. - fix: apply automatic fixes (equivalent to --fix). - select: list of rule codes to select (e.g., ["F401", "E501"]). - ignore: list of rule codes to ignore. - config: inline configuration or path to a ruff config file (passed via --config). - preview: enable preview mode (--preview). - quiet: reduce output verbosity (--quiet). - extra_args: additional raw arguments passed through as-is.

Run 'ruff format' to format Python files. - paths: files or directories to format; defaults to current directory if not provided. - check: don't write changes; exit non-zero if files would be reformatted (--check). - diff: print the diff of changes that would be made (--diff). - quiet: reduce output verbosity (--quiet). - preview: enable preview mode (--preview). - extra_args: additional raw arguments passed through as-is.

Get the Ruff version string.

Scan a directory or file path for secrets and return findings as a JSON list.

Scan a git repository for secrets and return findings as a JSON list.

Scan the provided string for secrets via stdin and return findings as a JSON list.

Create a scatter plot and return the saved PNG image.

Compute the series expansion of an expression. Args: expr: Expression, e.g., "1/cos(x)". var: Variable name, e.g., "x". point: Expansion point (default "0"). order: Series order (e.g., 6 means terms up to O((x - point)**6)). Returns: The series as a string (including the big-O term).

Preview or perform rewrite using an AST pattern. - pattern: match pattern - rewrite: replacement template - lang: language for parsing - path: directory or file to operate on - dry_run: if True, only previews changes (does not write). If False, attempts to write changes.

Search for AST pattern within a path. - pattern: ast-grep pattern, e.g., 'var code = $PATTERN' - lang: language for parsing (e.g., ts, js, py, rs, etc.) - path: directory or file to search (default '.') - json_output: output matches in JSON format

Get ast-grep version string.

Lint one or more shell script files and return structured diagnostics. - paths: list of file paths to analyze - shell: e.g. "bash", "sh", "dash", "ksh" - severity: minimum severity to display: "error", "warning", "info", or "style" - exclude_codes: list of rule codes (e.g. ["SC2086","SC2046"]) to suppress - include_dirs: list of directories to search for 'source'd files (-I) - allow_external_sources: if True, allow external sources (-x)

Lint a shell script provided as text and return structured diagnostics. - script: the shell script contents - shell: e.g. "bash", "sh", "dash", "ksh" - severity: minimum severity to display: "error", "warning", "info", or "style" - exclude_codes: list of rule codes (e.g. ["SC2086","SC2046"]) to suppress - include_dirs: list of directories to search for 'source'd files (-I) - allow_external_sources: if True, allow external sources (-x)

Return ShellCheck version string (shellcheck --version).

Simplify a symbolic expression. Args: expr: A SymPy-parsable expression string, e.g., "sin(x)**2 + cos(x)**2". Returns: The simplified expression as a string.

Solve expr = 0 for a single variable using solveset. Args: expr: Expression to solve, e.g., "x**2 - 4". var: Variable name to solve for, e.g., "x". Returns: A string representation of the solution set (can be a set, FiniteSet, ConditionSet, etc.).

Compute Spanish-specific readability metrics.

Split a PDF into single-page PDFs. - output_dir: directory to write outputs (created if needed). Defaults to ./split_<basename>. Returns a list of absolute paths to the created PDFs.

Optimize a single SVG file with SVGO. Args: input_path: Path to an existing .svg file (read-only). output_path: Optional path for the optimized SVG. If omitted, writes to ./out/<name>.min.svg. multipass: If true, run SVGO in multipass mode. Returns: Absolute path to the optimized SVG file.

Optimize an SVG provided as a string. Args: svg: SVG content as a UTF-8 string. multipass: If true, run SVGO in multipass mode. Returns: Optimized SVG content as a string.

Convert an expression to LaTeX. Args: expr: Expression string, e.g., "sin(x)/x". Returns: The LaTeX string.

Transcribe speech in an audio or video file to text, offline. Accepts common formats: m4a (voice memos), mp3, wav, flac, ogg, opus, aac, mp4, mov, mkv, webm. Multilingual: the language is detected unless given. Args: audio_path: Path to the file (read-only). language: Optional language code hint, e.g. "en", "es", "hi". task: "transcribe" (keep the spoken language) or "translate" (into English). word_timestamps: Also return per-word timings. Returns: {"text", "language", "language_probability", "duration_seconds", "segments": [{start, end, text}]}, with "warning" first when the l…

Translate a transcript to a target language using YouTube's translation feature. Args: video_id: The YouTube video ID. source_languages: List of language codes to search for the original transcript (e.g., ["en"]). target_language: Language code to translate to (e.g., "de"). preserve_formatting: If True, keep basic HTML formatting. Returns: A dict of the translated transcript metadata and snippets.

Trim a local video between start and end (in seconds) and write it to an output file. - input_path: path to the source video (must exist). - start: start time in seconds (optional). - end: end time in seconds (optional). - output_path: output file name (defaults to <input_stem>_trimmed.mp4 in the tool's own folder; the folder the input is in is usually read-only). - codec: video codec to use (default: libx264). - fps: output frames per second (optional; uses source FPS if None). - audio: whether to include audio (default: True). - audio_codec: audio codec (default: aac). Ignored if audio=Fa…

Validate a JSON instance against a JSON Schema. Args: instance: The JSON data to validate (any JSON-serializable value). schema: The JSON Schema to validate against. draft: Which JSON Schema draft to use. One of: "auto" (default), "2020-12", "2019-09", "7", "6", "4", "3". format_checks: If True, enable format checking (requires extras installed). max_errors: If set, stop collecting further errors after this many. Returns: A dict with: - valid: bool indicating whether the instance is valid. - error_count: number of validation errors found. - errors: list of er…

Return the installed Trafilatura version.

Read basic metadata (duration, fps, size) from a local video.

Compute the Wiener Sachtextformel (German) readability score. - variant: select formula variant (integer as per textstat docs) - lang: language code (defaults to de_DE)

Convert content from one format to another using yq. Equivalent to running '.' as the expression. - content: the input document(s) as a string. - input_format: parse format (e.g., json, yaml, xml, ini, toml, hcl, csv, tsv, props). - output_format: output format (e.g., json, yaml, xml, ini, toml, hcl, csv, tsv, props). - pretty_print: pretty print output when applicable (e.g., yaml).

Evaluate a yq expression against one or more files. - expression: jq-like expression to run. - files: list of file paths to process; if omitted, reads from STDIN (use yq_eval_string instead). - input_format: one of [auto|yaml|json|kyaml|props|csv|tsv|xml|base64|uri|toml|hcl|lua|ini]. - output_format: one of [auto|yaml|json|kyaml|props|csv|tsv|xml|base64|uri|toml|hcl|shell|lua|ini]. - pretty_print: pretty print output (shorthand for '... style = ""' for yaml). - unwrap_scalar: print just the scalar value with no quotes/colors/comments (equivalent to -r). - null_input: do not read input, just e…

Evaluate a yq expression on in-memory content provided via STDIN. - expression: jq-like expression to run. - content: the input document(s) as a string. - input_format: parse format (yaml/json/xml/etc.). Defaults to auto if not set. - output_format: output format (yaml/json/xml/etc.). Defaults to auto if not set. - pretty_print: pretty print output (for yaml pretty-print). - unwrap_scalar: print just the scalar value without quotes/colors/comments (-r).

Get yq version string.

Download media or extract info using yt-dlp. - urls: One or more URLs to process - output_dir: Directory to place outputs (default: current directory) - output_template: Custom output template (default: "%(title)s [%(id)s].%(ext)s") - format: Optional format selector (e.g., "bestvideo+bestaudio/best") - simulate: If True, do not download; only extract info and compute filenames - flat_playlist: If True, do not resolve playlist entries fully - restrict_filenames: If True, restrict filenames to ASCII and avoid special chars - verbose: If True, emit verbose logs

List all supported extractor names.

Get yt-dlp version string.