anbeime/skill / skills/handraw-style/handraw-style/skills/article-illustration-planner
article-illustration-planner
Verified深度分析文章脉络,精准规划插图点位与视觉意象。对知识类/教程类文章默认启动“认知降负与高保真图替繁文机制”,严禁删除公式与量化指标,强制协同 170 种排版图型(IG-*信息图/SC-*社媒卡)与 365 种视觉风格/36 种经典配色,杜绝微缩杂物摆拍;对非知识类文章保持叙事氛围与意境。严格执行双向间距(400~1200字)与全篇章节均衡覆盖,支持自主生图或全自动插图回填排版。
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 16.1 KB |
| scripts/insert_illustrations.py | script | 7.8 KB |
scripts/insert_illustrations.py
View on GitHub#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
insert_illustrations.py - Deterministic Markdown Article Illustration Inserter & Text Pruner
Part of the article-illustration-planner Skill.
Supports:
1. Pure insertion: Insert image embeds at designated anchor points (after/before heading or paragraph).
2. High-fidelity text replacement & pruning: Replace verbose/complex text blocks with image embeds + simplified summary
to reduce cognitive load in knowledge articles while enforcing zero knowledge loss on core formulas and metrics.
"""
import argparse
import json
import os
import re
import sys
from pathlib import Path
def audit_knowledge_fidelity(replace_text: str, simplified_prose: str, index: int | str):
"""
Audits whether critical formulas or numbers in replaced text were preserved in simplified text.
Warns if high-value knowledge items (LaTeX formulas, specific percentages, pricing, ratios) disappear.
"""
warnings = []
# Check LaTeX math formulas
if ("$$" in replace_text or "\\text" in replace_text or "\\times" in replace_text) and ("$$" not in simplified_prose and "\\text" not in simplified_prose and "×" not in simplified_prose and "*" not in simplified_prose):
warnings.append("LaTeX mathematical formula detected in replaced text but not in simplified prose.")
# Check key percentages
pcts = re.findall(r"\d+(?:\.\d+)?%", replace_text)
if pcts:
missing_pcts = [p for p in set(pcts) if p not in simplified_prose]
if missing_pcts and len(missing_pcts) == len(set(pcts)):
warnings.append(f"Quantitative metrics {missing_pcts} detected in replaced text; ensure diagram visualizes them.")
if warnings:
print(f"[FIDELITY AUDIT NOTICE - Illus {index}] " + " | ".join(warnings))
def insert_illustrations(
article_text: str,
manifest: list[dict],
) -> str:
"""
Inserts image embeds into article_text at anchor points or replaces text blocks.
Manifest schema:
[
{
"index": 1,
"anchor": "### 1. 概念起源",
"position": "after", # "after" or "before"
"image_path": "images/illus_01.webp",
"caption": "图1:概念起源与核心意象",
"alt": "插图1: 概念起源", # optional
"replace_text": "[被替换的原文字段]", # optional: for knowledge articles
"simplified_prose": "[精炼后的替代文本]" # optional: summary prose after image
}
]
"""
processed_text = article_text
# First pass: Handle explicit text replacements (Knowledge article cognitive pruning)
for item in manifest:
replace_text = item.get("replace_text", "").strip()
if not replace_text:
continue
image_path = item.get("image_path", "")
caption = item.get("caption", "")
alt = item.get("alt", caption or f"插图{item.get('index', '')}")
simplified_prose = item.get("simplified_prose", "").strip()
# Knowledge fidelity audit
audit_knowledge_fidelity(replace_text, simplified_prose, item.get("index", ""))
# Build replacement block
parts = [f"\n\n\n"]
if caption:
parts.append(f"*▲ {caption}*\n\n")
else:
parts.append("\n")
if simplified_prose:
parts.append(f"{simplified_prose}\n\n")
replacement_block = "".join(parts)
# Look for replace_text
if replace_text in processed_text:
processed_text = processed_text.replace(replace_text, replacement_block, 1)
item["_handled"] = True
print(f"[REPLACED] Successfully replaced complex text block with illustration {item.get('index', '')}")
else:
# Fuzzy match: try line-trimmed match
trimmed_target = "\n".join(line.strip() for line in replace_text.splitlines() if line.strip())
found = False
# Check paragraphs
for chunk in processed_text.split("\n\n"):
trimmed_chunk = "\n".join(l.strip() for l in chunk.splitlines() if l.strip())
if trimmed_target and trimmed_target in trimmed_chunk:
processed_text = processed_text.replace(chunk, replacement_block, 1)
item["_handled"] = True
found = True
print(f"[REPLACED-FUZZY] Replaced matched paragraph with illustration {item.get('index', '')}")
break
if not found:
sys.stderr.write(f"[WARNING] replace_text not found in article for illustration {item.get('index', '')}. Will fall back to anchor insertion.\n")
# Second pass: Handle pure anchor insertions for remaining items
lines = processed_text.splitlines(keepends=True)
for item in manifest:
if item.get("_handled"):
continue
anchor = item.get("anchor", "").strip()
if not anchor:
continue
position = item.get("position", "after").lower()
image_path = item.get("image_path", "")
caption = item.get("caption", "")
alt = item.get("alt", caption or f"插图{item.get('index', '')}")
simplified_prose = item.get("simplified_prose", "").strip()
# Build insertion block
parts = [f"\n\n"]
if caption:
parts.append(f"*▲ {caption}*\n\n")
else:
parts.append("\n")
if simplified_prose:
parts.append(f"{simplified_prose}\n\n")
insertion_block = "".join(parts)
# Find matching line
matched_idx = -1
for idx, line in enumerate(lines):
if anchor in line:
matched_idx = idx
break
if matched_idx != -1:
if position == "before":
lines.insert(matched_idx, insertion_block)
else:
lines.insert(matched_idx + 1, insertion_block)
print(f"[INSERTED] Inserted illustration {item.get('index', '')} {position} anchor: '{anchor}'")
else:
sys.stderr.write(f"[WARNING] Anchor not found in article: '{anchor}'. Appending illustration.\n")
lines.append(insertion_block)
return "".join(lines)
def main():
parser = argparse.ArgumentParser(description="Insert illustrations and prune complex text in Markdown article.")
parser.add_argument("--article", "-a", required=True, help="Path to input Markdown article.")
parser.add_argument("--manifest", "-m", help="Path to JSON manifest of illustrations.")
parser.add_argument("--manifest-json", help="Direct JSON string of manifest.")
parser.add_argument("--output", "-o", help="Path to output Markdown file. Defaults to [stem]_illustrated.md")
args = parser.parse_args()
article_path = Path(args.article)
if not article_path.exists():
sys.stderr.write(f"Error: Article file '{article_path}' does not exist.\n")
sys.exit(1)
article_text = article_path.read_text(encoding="utf-8")
manifest = []
if args.manifest:
manifest_file = Path(args.manifest)
if not manifest_file.exists():
sys.stderr.write(f"Error: Manifest file '{manifest_file}' does not exist.\n")
sys.exit(1)
manifest = json.loads(manifest_file.read_text(encoding="utf-8"))
elif args.manifest_json:
manifest = json.loads(args.manifest_json)
else:
sys.stderr.write("Error: Either --manifest or --manifest-json must be provided.\n")
sys.exit(1)
illustrated_text = insert_illustrations(article_text, manifest)
if args.output:
output_path = Path(args.output)
else:
output_path = article_path.parent / f"{article_path.stem}_illustrated{article_path.suffix}"
output_path.write_text(illustrated_text, encoding="utf-8")
print(f"[SUCCESS] Illustrated article written to: {output_path}")
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 197k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |