Campaign Report
Based on the project background, goals, implementation actions, and verified data, generate executive summaries, comprehensive reviews, poster copy, or social media reports. This is for project managers, marketing, and growth teams to use when reporting phase achievements or campaign results.
Unlabeled version · MEXC Design · Updated on July 22, 2026
Quick Start
1. Download the ZIP file; extraction is not required.
2. Open Claude Desktop → Customize → Skills → + → Upload a skill, and select this ZIP file.
3. Turn on the switch for this Skill in the list, and then describe your requirement as usual; Claude will call it automatically.
First, enable "Code execution and file creation" in Settings → Capabilities.
Workflow
1
Provide requirements
2
Understanding and Planning
3
Generate creative draft
4
Optimize and Finalize
5
Deliverables
Rules and Constraints
Do not fabricate, guess, or fill in missing data.
Percentages must specify the denominator and the time range.
Correlation must not be expressed as causation.
Sensitive information has been removed from the external version.
Full SKILL.md
Convert structured data from Lark Docs, Sheets, Wikis, and PDF attachments into high-information-density, visually clear HTML summary charts, and complete the full closed-loop process of "screenshot → upload to Lark → return link". It supports multi-data source fusion, period-over-period comparisons, and visualization of OKR objectives and progress.
Prerequisites
· lark-cli is installed and logged in
· Playwright and Chromium are already installed in the Python environment
· PDF data source also requires pymupdf
· Have your company logo materials ready
Required input
· Data source: Links or tokens of Feishu documents, spreadsheets, wikis, or PDF attachments
· Report Subject and Time Range
· Core metrics, target values, and comparison baselines that need to be highlighted
· Scenario Type: Weekly progress, health status inspections, business promotion, or OKR dashboards
· Risks and Action Items
· Output size, default 1800×1100 landscape double-column
Execution Flow
1. Data Acquisition: Use fetch and specify the v2 interface for documents, read tables using sheets, and for PDFs, download them first before extracting text with pymupdf.
2. Multi-source Fusion: When comparing multiple documents, they are read separately and then their fields are merged, with key extraction focusing on the current value, previous baseline, target value, achievement rate, month-on-month change, risks, and action items.
3. HTML Generation: Call the template to inject data, and align with the enterprise visual standards before use.
4. Screenshot: Open with Playwright at the target width, wait for rendering to complete, get the actual height, and take a full-page screenshot.
5. Upload to Feishu: Create a document and insert images, then return the document link and local HTML and PNG paths.
Output Contract
Lark Document Link — The address of the document with embedded images.
Local HTML — re-editable source file.
Local PNG — finished screenshot.
Data Source Section — Lists the data sources one by one at the bottom of the page.
Design Guidelines
· Page background #F3F4F7, theme color #0057FF, body text #000000, content block white background with 12px rounded corners.
· The top header bar has a blue background with white text. The enterprise logo should be placed in the top-left corner, with a width not exceeding 180px and a height not exceeding 48px.
· Semantic Status Colors: Red #E53935 for High Risk, Yellow #F59E0B for Medium Risk, Blue #0057FF for Normal Progress, Green #16A34A for Ahead of Plan.
· Unified font to Noto Sans SC: Main Title 60px/900, Project Title 26px/800 Blue, Body Text 16px/400 Black.
· 8pt grid, cards in the same row have equal width and height, with left and right edges strictly aligned.
· Structural Order: Title Bar → KPI Overview Bar → Two-column Main Body → Risks & Actions → Data Source Bar.
Execute Rules
· Document reading must specify the v2 interface; v1 is deprecated.
· Inserting images must use relative paths; absolute paths will trigger an "unsafe file path" error.
· Embedded tables in the document need to be read separately, as key metrics are often contained within them.
· Wait for the page to render completely before taking a full-page screenshot at its actual height.
· Do not use dark themes, glassmorphism, gradients, shadows, and extra decorative colors.
· Do not use outlining or repetitive drawing to simulate bolding; establish text hierarchy through font size, font weight, and negative space.
· Long text must wrap; no text should overflow the white block. When crowded, prioritize increasing the canvas height.
Call template
Use the visual report generation Skill to generate a progress overview image for this week based on the following Feishu document. Read the document and its embedded tables, extract the target values, current values, and week-over-week changes of core metrics, and output the HTML, generate a screenshot, and upload it to Feishu to return the link.
Example
Input:
Quick weekly update on mid-end technology progress generated based on a Feishu document.
Expected behavior:
1. Read the document and extract the embedded table tokens to read the OKR data.
2. Apply the weekly report template to generate HTML and write it locally.
3. Take a full-page screenshot at 1800px wide.
4. Create a Lark document, insert images using relative paths, and return the links and local paths.