Lab Report Writer

Lab Report Writer

Generate professional, low-plagiarism lab reports with proper structure, data analysis, and references.

What This Skill Does

1. Three input modes: From topic/description → full report; From raw data → analysis + report; From draft → improvement & formatting
2. Covers all report types: University course labs, scientific research, engineering tests, medical/material experiments
3. Auto-research: WebSearch for theoretical background, related work, and real citable references
4. Data processing: Tables, charts (line/scatter/bar with error bars), uncertainty propagation, least-squares fitting
5. Low plagiarism: Original writing throughout, proper paraphrasing, no copy-paste
6. Multiple output formats: docx (with tables/charts/formulas) or markdown, user's choice

Prerequisites

• Node.js 18+ (for docx generation via npm docx package)
• If docx output: docx npm package must be available (npm list -g docx or project-local)
• For charts: Python 3 with matplotlib, OR use ASCII/markdown tables as fallback

Level 2: Quick Start

Interactive Launch

When the user mentions writing a lab report, experiment report, or test report, start the interactive flow:

/lab-report

The skill will ask clarifying questions via AskUserQuestion before generating anything.

Level 3: Detailed Instructions

Step 1: Interactive Questionnaire

Use AskUserQuestion to collect the following information. Ask in 1-2 rounds (max 4 questions per round).

Round 1 — Core Parameters

Round 2 — Content Details (adapt based on Round 1 answers)

Step 2: Load Grading Criteria (If Provided)

If the user provides a grading rubric file path (e.g., 成绩评定表.docx):

1. Read the file using textutil -convert txt -stdout (macOS) or Python
2. Extract scoring dimensions and weights
3. Align the report structure and content depth to maximize score on each dimension
4. Store extracted criteria in memory for self-check at the end

Step 3: Research Phase

Execute WebSearch queries to gather:

1. Theoretical background: Core principles, equations, mechanisms
2. Related work: Key papers, review articles, textbook references
3. Reference literature: Real, verifiable sources with complete bibliographic info

Search strategy:

• Start broad: "[topic] 实验原理 综述" or "[topic] experiment principle review"
• Then specific: "[topic] 数据处理 误差分析" or "[topic] data analysis uncertainty"
• For references: "[author] [key term]" IEEE/Sensors/Review to find specific papers
• Run 10-20 searches for thorough coverage; 5-8 for shorter reports

Critical: Record every search result that yields a usable reference. Do NOT fabricate references.

Step 4: Section Template Selection

Based on report type, select the appropriate section template from Section Templates below.

Step 5: Data Processing (If Applicable)

If the user provides experimental data:

1. Read data: CSV via Read, Excel via Python, or JSON directly
2. Generate tables: Format as markdown tables in the report; for docx, use Table objects
3. Compute statistics: Mean, standard deviation, relative error
4. Uncertainty propagation: Apply formula Δy = √(Σ(∂y/∂xi · Δxi)²)
5. Curve fitting: Linear least-squares y = ax + b with R², or polynomial if needed
6. Charts: Generate using matplotlib (save as PNG) or describe as ASCII for markdown

For data processing code, use Python via Bash:

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
# ... analysis code ...

Step 6: Write the Report

Write section by section, following these principles:

Low-Plagiarism Writing Rules

1. Original phrasing: Never copy sentences from sources. Read, understand, rephrase in your own words.
2. Structure variation: Don't follow the exact paragraph structure of any single source.
3. Specific details: Include concrete numbers, formulas, and technical terms — generic text is more likely to match existing content.
4. Personal analysis: The "discussion/analysis" section should reflect independent reasoning, not paraphrased conclusions.
5. Chinese-English mixing: For Chinese reports, mixing technical terms in English (with Chinese explanation) naturally lowers similarity scores.

Section Writing Order

Recommended order for efficiency:

1. Write the body sections first (principle, procedure, data, analysis)
2. Then introduction and conclusion (these frame the body)
3. Then abstract (summarizes everything)
4. Finally compile references

Step 7: Format and Generate Output

For docx output:

Use Node.js with the docx npm package. Generate using a script:

const docx = require("docx");
const { Document, Packer, Paragraph, TextRun, AlignmentType, Table, ... } = docx;
// Build document with proper formatting
// Font: SimSun (宋体) body, SimHei (黑体) headings, KaiTi (楷体) abstract
// Line spacing: 1.5x (360 twips)
// Margins: 2.54cm top/bottom, 3.17cm left/right
// First-line indent: 2 characters (~480 twips)

Find the docx module path:

# Check global install
npm list -g docx 2>/dev/null
# Use NODE_PATH when running
NODE_PATH=$(npm root -g) node generate-report.js

For markdown output:

Write directly with Write tool. Include:

• YAML frontmatter with title, author, date
• Proper heading hierarchy
• Markdown tables for data
• Math notation with $...$ or $$...$$
• Image references for charts

Step 8: Self-Check Against Grading Criteria

If grading criteria were loaded, verify:

Step 9: Deliver

1. Save output file to the user's specified location (default: ~/Desktop/)
2. Report word count (Chinese characters + English terms)
3. List any manual edits needed (e.g., author name, school info)
4. Clean up temporary files (generation scripts, intermediate data)

Section Templates

Template A: University Course Lab Report

Template B: Scientific Research Report

Template C: Engineering Test Report

Template D: Medical/Material Experiment Report

Reference Format Guide (GB/T 7714)

Journal Article

[1] Author1, Author2, Author3. Title[J]. Journal Name, Year, Volume(Issue): Pages.

Example:

[1] Smith C S. Piezoresistance effect in germanium and silicon[J]. Physical Review, 1954, 94(1): 42-49.

Book/Monograph

[2] Author. Title[M]. Edition. Place: Publisher, Year: Pages.

Conference Paper

[3] Author1, Author2. Title[C]//Conference Name. Place: Publisher, Year: Pages.

Technical Report

[4] Author/Organization. Title[R]. Place: Institution, Year.

Online Resource

[5] Author. Title[EB/OL]. (Date)[Cite Date]. URL.

Data Sheet / Standard

[6] Organization. Title[Z]. Place: Organization, Year.

Error Analysis Formulas

Direct Measurement Uncertainty

Type A (statistical): $$u_A = \sqrt{\frac{\sum_{i=1}^{n}(x_i - \bar{x})^2}{n(n-1)}}$$

Type B (instrument): $$u_B = \frac{\Delta_{instr}}{\sqrt{3}}$$ (uniform distribution)

Combined standard uncertainty: $$u_c = \sqrt{u_A^2 + u_B^2}$$

Expanded uncertainty (k=2 for 95% confidence): $$U = k \cdot u_c = 2u_c$$

Indirect Measurement (Propagation)

If $y = f(x_1, x_2, ..., x_n)$: $$u_c(y) = \sqrt{\sum_{i=1}^{n}\left(\frac{\partial f}{\partial x_i}\right)^2 u_c^2(x_i)}$$

Linear Least-Squares Fit

For $y = ax + b$: $$a = \frac{n\sum x_iy_i - \sum x_i\sum y_i}{n\sum x_i^2 - (\sum x_i)^2}$$ $$b = \frac{\sum y_i - a\sum x_i}{n}$$ $$R^2 = \frac{(n\sum x_iy_i - \sum x_i\sum y_i)^2}{[n\sum x_i^2 - (\sum x_i)^2][n\sum y_i^2 - (\sum y_i)^2]}$$

Relative Error

$$\delta = \frac{|x_{measured} - x_{true}|}{x_{true}} \times 100%$$

Chart Generation Guidelines

Using Python + matplotlib

import matplotlib
matplotlib.use('Agg')  # Non-interactive backend
import matplotlib.pyplot as plt
import numpy as np

# Configure for Chinese labels
plt.rcParams['font.sans-serif'] = ['Arial Unicode MS', 'SimHei', 'Heiti TC']
plt.rcParams['axes.unicode_minus'] = False

# Example: Line chart with error bars
fig, ax = plt.subplots(figsize=(8, 5))
x = np.array([1, 2, 3, 4, 5])
y = np.array([2.1, 3.8, 6.2, 7.9, 10.1])
y_err = np.array([0.2, 0.3, 0.2, 0.4, 0.3])

ax.errorbar(x, y, yerr=y_err, fmt='o-', capsize=4, label='Measured')
ax.set_xlabel('Independent Variable')
ax.set_ylabel('Dependent Variable')
ax.set_title('Experiment Results')
ax.legend()
ax.grid(True, alpha=0.3)

plt.tight_layout()
plt.savefig('chart.png', dpi=200)
print("Chart saved: chart.png")

Chart Types by Data

Troubleshooting

Issue: docx npm module not found

Solution: Find the global install path and set NODE_PATH:

npm list -g docx  # Find install location
NODE_PATH=$(npm root -g) node generate-report.js

Issue: Chinese fonts not rendering in docx

Solution: Use standard font names: SimSun (宋体), SimHei (黑体), KaiTi (楷体). On macOS, these may render as fallback — the document will display correctly on Windows with Office.

Issue: matplotlib Chinese characters show as boxes

Solution: Add font configuration:

plt.rcParams['font.sans-serif'] = ['Arial Unicode MS', 'SimHei', 'Heiti TC']

Issue: WebSearch returns no results

Solution: Try English queries, simplify terms, search for specific author+keyword combinations. Fall back to general knowledge if needed (mark references as "[待核实]").

Issue: Data file format not recognized

Solution: For Excel (.xlsx), use Python with openpyxl or pandas. For .csv, use Read tool directly.

Related Skills

• kimi-webbridge: For web research when WebSearch is disabled or insufficient
• ui-styling: For chart styling and visual design
• design: For report cover page or presentation-quality figures