Content goes here...
\", \"keyword\": \"mcp\"}'\nspeedgrapher call vale '{\"text\": \"This is very unique.\"}'\n\n# Pipe text directly into tool execution via stdin\ncat draft.md | speedgrapher call slop\n```\n\n### MCP Server Execution\n\nSpeedgrapher can run as a standard stdio server or as a network-accessible streamable HTTP service:\n\n```bash\n# Run stdio MCP server (standard for MCP clients like Claude Code or Gemini)\nspeedgrapher mcp\n\n# Run streamable HTTP MCP server\nspeedgrapher mcp --listen=:8080\n```\n\n---\n\n## Available MCP Tools\n\n| Tool | Summary |\n| :--- | :--- |\n| [`fog`](#fog) | Calculates the Gunning Fog Index to estimate text readability and audience classification. |\n| [`slop`](#slop) | Multi-metric heuristic analyzer calculating an AI cliché / slop score (0–100). |\n| [`analyze_seo`](#analyze_seo) | Comprehensive 7-point technical SEO audit for published URLs or Hugo Markdown drafts. |\n| [`vale`](#vale) | Automated Vale static analysis engine verifying editorial style, grammar, and voice. |\n\n### Tool Breakdown & Behaviors\n\n#### `fog`\n- **Parameters**: \n - `text` (string, optional): Text to analyze for readability. Must contain at least one sentence.\n - `path` (string, optional): Path to a file containing text to analyze. Resolved against `WorkspaceDir`.\n- **Behavior**:\n - Calculates the Gunning Fog Index using average sentence length (ASL) and percentage of complex words (PCW, words with 3+ syllables):\n $$\\text{Fog Index} = 0.4 \\times \\left( \\frac{\\text{words}}{\\text{sentences}} + 100 \\times \\frac{\\text{complex words}}{\\text{words}} \\right)$$\n - Returns classification:\n - **Simplistic** ($< 9$): Accessible for elementary reading levels.\n - **General Audiences** ($9 \\le \\text{Index} < 13$): Clear and accessible for most readers.\n - **Professional Audiences** ($13 \\le \\text{Index} < 18$): Ideal for technical and engineering blogs.\n - **Hard to Read** ($18 \\le \\text{Index} < 22$): Requires significant cognitive effort.\n - **Unreadable** ($\\ge 22$): Likely incomprehensible to general technical audiences.\n - Returns structured metrics: `fog_index`, `classification`, `total_words`, `total_sentences`, `average_sentence_length`, `percentage_complex_words`, `complex_words`.\n\n#### `slop`\n- **Parameters**:\n - `text` (string, optional): Text to analyze for AI-generated clichés and structural tropes.\n - `path` (string, optional): Path to a file containing text to analyze.\n- **Behavior**:\n - Computes a weighted overall score ($0$ to $100$) across 5 calibrated analytical dimensions:\n 1. **Structural Clichés (40% weight)**: Scans for distinct LLM rhetorical patterns (e.g., *\"It's not X — it's Y\"*, *\"Not X. Not Y. Just Z.\"*, *\"The result? Devastating.\"*, *\"Here's the kicker\"*, *\"Delve into the tapestry\"*, em-dash clusters).\n 2. **Lexical Slop (25% weight)**: Detects overused AI buzzwords (*delve*, *tapestry*, *landscape*, *nuance*, *testament*, *beacon*, *catalyst*, *paradigm*, *robust*, *seamless*, *transformative*, *quietly*, *deeply*, *fundamentally*).\n 3. **Filler Words (15% weight)**: Measures stop-word and filler ratios against natural human writing distributions.\n 4. **Rhythm Variance (15% weight)**: Evaluates sentence length coefficient of variation (CV) to penalize uniform, monotonous sentence cadences.\n 5. **Syntactic Voice (5% weight)**: Measures pronoun-to-noun balance via part-of-speech (POS) tagging.\n\n#### `analyze_seo`\n- **Parameters**:\n - `url` (string, optional): Full URL of the live webpage to audit.\n - `html` (string, optional): Raw HTML string or Hugo Markdown with YAML front matter.\n - `keyword` (string, optional): Target keyword to verify across title, description, and headings.\n- **Behavior**:\n - Runs a 7-point technical SEO inspection:\n 1. **Title Tag**: Checks presence, optimal character length ($30$–$60$ chars), and keyword placement.\n 2. **Meta Description**: Checks presence, optimal length ($120$–$160$ chars), and keyword placement.\n 3. **H1 Tag**: Enforces single H1 presence and keyword inclusion.\n 4. **Image Alt Text**: Detects missing or empty `alt` attributes on all `