What problem does it solve? Large-scale terminology changes in technical documents (SRS, design docs, API specs) are error-prone: missed occurrences, corrupted mermaid diagrams, broken markdown tables, and inconsistent Chinese-English mixed terms. This Skill provides a controlled five-phase workflow that audits every occurrence first, grounds proposals in real-world product references, requires explicit approval before edits, and verifies results with grep checks. ## Core Features & Use Cases - Five-phase revision workflow: Audit with multi-pattern grep, reference canonical products (VS Code, JetBrains Rider), propose a BEFORE/AFTER mapping table, execute replacements in dependency order, and verify zero-count of old terms. - Breakage recovery patterns: Detects and repairs replace_all corruption of mermaid ER diagrams (||--o{), falls back to line-targeted sed when the patch tool fails on SVG text with escaped quotes, and cleans up double-space artifacts after replacement. - Use Case: You need to rename "Dock 区域" to "停靠区" across a 2000-line design document. The Skill maps all occurrences, proposes a terminology table aligned with VS Code conventions, waits for your approval, then executes replacements from longest compound phrases to standalone terms and confirms old terms are gone. ## Quick Start Ask the agent to audit and rename a term across your markdown document, for example: "Replace all occurrences of 'Dock Panel' with 'Panel' in design.md and show me the mapping table before making changes."
