Convert documents to PDF and render HTML templates to images. Use html_to_pdf for reports and invoices, render_image for social cards and banners, convert_office_document for DOCX/XLSX/PPTX.
Four tools with complete input schemas and solid descriptions (avg 180 chars). All parameters typed and described. Tool names are action-oriented (html_to_pdf, render_image, convert_office_document, merge_pdfs). Output schemas not explicitly documented in source. No error handling guidance visible. No tool annotations (readOnlyHint, destructiveHint). Descriptions are clear but lack dependency hints and recovery guidance. Parameters have good constraints (defaults, enums for format). Security posture is strong: no secrets in params, read-only operations only.
Convert an office document to PDF. Accepts DOCX, XLSX, PPTX, ODT, RTF, CSV and similar formats. `filename` must carry the real extension — it is how the converter picks a filter. `content_base64` is the raw file, base64 encoded. Returns the PDF as a base64 data URI.
Convert an HTML document to PDF. Use this when the user wants a report, invoice, or formatted document as a PDF file. Supply a complete HTML document including any inline CSS. Returns the PDF as a base64 data URI.
Merge two or more PDFs into a single document, in the order given. Each entry in `files_base64` is one base64-encoded PDF. Returns the merged PDF as a base64 data URI.
Render an HTML template to an image. Use this for social share images (Open Graph cards), banners, certificates, or any picture built from text and layout. The template is HTML with Jinja2 placeholders such as {{ title }}; `data` supplies those values. Default size 1200x630 is the standard Open Graph card. Returns the image as a base64 data URI.
Output schemas not documented. LLMs cannot plan downstream operations or extract fields from base64 data URIs without knowing structure.
No error handling guidance. Descriptions lack recovery hints (e.g., 'If conversion fails, check HTML syntax' or 'Unsupported file format, try DOCX instead').
Tool annotations missing. No readOnlyHint, destructiveHint, or idempotentHint declared, limiting agent planning and safety reasoning.
render_image description lacks guidance on when to use vs html_to_pdf. Both produce visual output; LLMs may conflate them.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 66 | 2026-07-28+ | v2 |
merge_pdfs accepts array of base64 strings but no limit documented. LLMs could pass 1000+ files, causing memory/timeout issues.