docs-router

# Daymade document router Choose the one specialist that owns the requested result. First find the path of **this loaded** `docs-router/SKILL.md` in the active skill catalog or the invocation that opened it. Resolve that path to its canonical location (follow symlinks), verify that it exists under a `docs-router` directory, and take that directory's parent as the suite root. The paths in the table are relative to this router's directory; resolve the selected path and verify that it points to a direct sibling `SKILL.md` inside that suite root. Never anchor resolution to the task's working directory or an old installed copy. Claude Code may expand `${CLAUDE_PLUGIN_ROOT}` as a shortcut, but the router must work when that variable is absent, as in Codex. If this router's own active path cannot be identified unambiguously, stop instead of guessing a copy. Read the selected child `SKILL.md` **in full** before acting; if a read is truncated, continue from the omitted part. Then load every reference that the child requires for this branch of work. The leaf skills are intentionally absent from automatic discovery, so open their files directly rather than invoking them through the Skill tool. A user can still invoke any leaf's original `/daymade-docs:<leaf>` command manually. | Requested result | Read this exact file | |---|---| | Convert a DOCX, PDF, PPTX, or saved HTML/HTM to Markdown; parse Word or extract document text and images into Markdown | `../doc-to-markdown/SKILL.md` | | Turn Markdown into a printable PDF; adapt a manufacturer PDF manual into a concise Chinese illustrated recipient guide | `../pdf-creator/SKILL.md` | | Create or format a Word `.docx`; export or repair an **existing Word/WPS manuscript** as PDF | `../docx-creator/SKILL.md` | | Turn a PDF into a self-contained, image-faithful HTML reading page, optionally translating it while keeping figures and charts | `../pdf-to-html/SKILL.md` | | Extract Mermaid diagrams from Markdown or render Mermaid as PNG | `../mermaid-tools/SKILL.md` | | Create a professionally formatted Excel workbook, parse a complex investment-bank `.xlsm` model, or control Excel on macOS | `../excel-automation/SKILL.md` | | Prepare printable ID/passport copies from existing photos, turn paper photos into a scanned PDF, replace scan pages, or make an unsigned digital document look hand-signed and scanned | `../photo-to-scanned-pdf/SKILL.md` | | Extract Word/WPS DOCX comments or tracked changes; compare explicitly selected DOCX versions, using the verified sent baseline for counterpart feedback, including files with no review traces | `../read-docx-review/SKILL.md` | | Check documentation impact after an authorized code, script, config, environment, path, deployment, auth, test, or procedure change; or consolidate redundant docs | `../docs-cleaner/SKILL.md` | Decide by the **input and requested output**, not a shared word such as “PDF” or “document.” For example, Markdown → PDF selects `pdf-creator`; an existing Word manuscript → PDF selects `docx-creator`; PDF → Markdown selects `doc-to-markdown`; PDF → HTML selects `pdf-to-html`. If the request spans two results, read each relevant child before its stage. For a manufacturer manual → Chinese illustrated guide (图文操作指南), select `pdf-creator` and load its conditional illustrated-manual reference before drafting. Use the sibling resolution above for any source-text or image extraction stage; keep ordinary PDF reading/editing with its existing owner. For website → Markdown, acquire the authorized pages through the owning web workflow first, then select `doc-to-markdown` for the saved HTML. That conversion branch does not crawl a site or certify presentation in a recipient's reader. Do not select a Daymade child for a new slide deck or generic PPT work. Use `deck-creator` for new presentation creation only when it is installed. If it is unavailable, this suite has no automatic current PPT builder; external users may explicitly invoke the retired `/daymade-docs:ppt-creator` for its compatibility notice only, not PPT generation. General PDF reading/editing, ordinary spreadsheet analysis, and other tasks without a matching row remain with their owning skills or normal workflow. If the result is unclear, inspect the actual file/task or ask for the missing input or output; do not guess a Daymade child from an extension alone. After an authorized code, script, config, environment, path, deployment, auth, test, or procedure change, select `docs-cleaner` for a scoped documentation impact check even if the user did not mention docs. Read its full instructions at that stage; do not replace the primary task's specialist or scan unrelated documentation. Explicit documentation impact and cleanup requests also route there. A read-only investigation does not acquire write authorization through this router.