ZJU-REAL/Easel / skills/builtin/skill-social-performance-review
skill-social-performance-review
Verified生成月度社媒效果复盘报告,分析小红书、抖音、B站、微博等平台的内容表现,输出下月可执行建议。 当用户说"月度复盘""效果复盘""这个月表现""内容复盘""运营总结""下月建议""月报"时使用。 和 skill-publish-analytics 的区别:analytics 从发布日志做四维归因,本 SKILL 做跨平台组合级月度复盘。
| File | Kind | Size |
|---|---|---|
| EASEL-META.md | markdown | 484 B |
| SKILL.md | skill md | 8.4 KB |
| references/analysis-framework.md | markdown | 3.8 KB |
| references/benchmarks.md | markdown | 7.5 KB |
| references/report-template.md | markdown | 2.1 KB |
| scripts/review.py | script | 8.1 KB |
scripts/review.py
View on GitHub#!/usr/bin/env python3
"""review.py — skill-social-performance-review 的确定性月度复盘计算.
把"逐帖评分 / 互动率 / Top-Bottom 排名 / 环比 / 加权内部评分"从 LLM
心算固化为代码。LLM 负责写洞察、原因诊断和下月建议。
子命令:
score 读入标准化的月度帖子 JSON,输出逐帖互动率、Top/Bottom、
支柱/格式聚合、环比、加权内部评分(1-10)
输入 JSON(--input)结构见 SKILL.md「确定性计算」节。缺字段自动降级,
不中断计算,并在 warnings 里说明。
"""
from __future__ import annotations
import argparse
import json
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).resolve().parents[3] / "shared" / "scripts"))
import social_stats as ss # noqa: E402
# 内部评分维度权重(与 SKILL.md / analysis-framework.md 一致)
SCORE_WEIGHTS = {
"engagement_vs_benchmark": 0.25,
"follower_growth": 0.20,
"top_post": 0.20,
"schedule_execution": 0.15,
"reach_trend": 0.20,
}
def load_input(path: Path) -> dict:
if not path.exists():
sys.exit(f"错误:未找到输入文件 {path}。请提供标准化的月度帖子 JSON。")
try:
with open(path, encoding="utf-8") as f:
return json.load(f)
except (OSError, json.JSONDecodeError) as e:
sys.exit(f"错误:无法读取 {path}:{e}")
def _denominator(post):
"""互动率基数:优先 reach,其次 impressions,再 views。返回 (值, 名称)。"""
for key in ("reach", "impressions", "views"):
v = post.get(key)
if v:
return v, key
return None, None
def _interactions(post):
"""互动总量 = likes + comments + saves + shares(缺失当 0,全缺返回 None)。"""
parts = [post.get(k) for k in ("likes", "comments", "saves", "shares")]
if all(p is None for p in parts):
return None
return sum(p or 0 for p in parts)
def score_posts(data):
posts = data.get("posts", [])
warnings = []
if not posts:
sys.exit("错误:输入中 posts 为空,无法复盘。")
# 逐帖:互动率、互动综合分
per_post = []
for p in posts:
inter = _interactions(p)
denom, denom_name = _denominator(p)
er = ss.engagement_rate(inter, denom, as_percent=True)
row = {
"title": p.get("title"), "date": p.get("date"),
"type": p.get("type"), "pillar": p.get("pillar"),
"reach": p.get("reach"), "impressions": p.get("impressions"),
"views": p.get("views"),
"likes": p.get("likes"), "comments": p.get("comments"),
"saves": p.get("saves"), "shares": p.get("shares"),
"interactions": inter,
"engagement_rate_pct": er, "er_denominator": denom_name,
"engagement_score": ss.engagement_score(
views=p.get("views"), likes=p.get("likes"),
comments=p.get("comments"), shares=p.get("shares")),
}
per_post.append(row)
# 排名:小红书优先看收藏,否则看互动率
platform = (data.get("platform") or "").lower()
if platform in ("xiaohongshu", "小红书", "xhs"):
rank_key, rank_name = (lambda r: r.get("saves") or -1), "saves"
else:
rank_key, rank_name = (lambda r: r.get("engagement_rate_pct") or -1), \
"engagement_rate_pct"
ranked = sorted(per_post, key=rank_key, reverse=True)
top3 = ranked[:3]
bottom3 = ranked[-3:][::-1] if len(ranked) > 3 else []
# 账号快照聚合
avg_er = ss.mean([r["engagement_rate_pct"] for r in per_post])
total_reach = sum(ss.clean(r["reach"] for r in per_post)) or None
snapshot = {
"post_count": len(per_post),
"avg_engagement_rate_pct": avg_er,
"median_engagement_rate_pct": ss.median(
[r["engagement_rate_pct"] for r in per_post]),
"total_reach": total_reach,
"total_interactions": sum(ss.clean(r["interactions"] for r in per_post)) or None,
"followers": data.get("followers"),
}
# 支柱 / 格式聚合
def agg_group(key):
groups = {}
for r in per_post:
groups.setdefault(r.get(key) or "未分类", []).append(r)
return {g: {"count": len(rs),
"avg_engagement_rate_pct": ss.mean(
[x["engagement_rate_pct"] for x in rs]),
"avg_reach": ss.mean([x["reach"] for x in rs]),
"avg_saves": ss.mean([x["saves"] for x in rs])}
for g, rs in groups.items()}
by_pillar = agg_group("pillar")
by_format = agg_group("type")
# 环比(vs previous)
prev = data.get("previous") or {}
mom = {
"engagement_rate_pct_change": ss.pct_change(
avg_er, prev.get("avg_engagement_rate_pct")),
"reach_change_pct": ss.pct_change(total_reach, prev.get("reach")),
"followers_change": data.get("followers_change"),
}
internal_score = _internal_score(data, per_post, top3, avg_er, mom, warnings)
if len(per_post) < 5:
warnings.append(ss.sample_warning(len(per_post), 5, "本月帖子"))
warnings = [w for w in warnings if w]
return {
"month": data.get("month"), "platform": data.get("platform"),
"rank_by": rank_name,
"account_snapshot": snapshot,
"per_post": per_post,
"top3": top3, "bottom3": bottom3,
"by_pillar": by_pillar, "by_format": by_format,
"month_over_month": mom,
"internal_score": internal_score,
"warnings": warnings,
}
def _internal_score(data, per_post, top3, avg_er, mom, warnings):
"""加权内部评分 1-10。缺失维度剔除并重新归一化。"""
comps = {}
# 互动率 vs 基准:avg_er / benchmark_avg 映射到 0-10(达标=6)
bench = (data.get("benchmark") or {}).get("engagement_rate_avg")
if avg_er is not None and bench:
ratio = avg_er / (bench * 100 if bench < 1 else bench)
comps["engagement_vs_benchmark"] = ss.scale_to_range(ratio, 0.5, 1.5, 0, 10)
else:
warnings.append("内部评分:缺基准或互动率,跳过'互动率vs基准'维度")
# 粉丝增长趋势:followers_change 相对基数映射
fc = data.get("followers_change")
followers = data.get("followers")
if fc is not None and followers:
growth_pct = ss.pct_change(followers, followers - fc) # 近似月增长率%
comps["follower_growth"] = ss.scale_to_range(growth_pct, -2, 5, 0, 10)
elif fc is not None:
comps["follower_growth"] = 6.0 if fc > 0 else (3.0 if fc == 0 else 1.0)
# 最佳帖子表现:Top1 互动率相对全月均值
if top3 and avg_er:
t1 = top3[0].get("engagement_rate_pct")
if t1 is not None:
comps["top_post"] = ss.scale_to_range(
ss.safe_div(t1, avg_er), 1.0, 3.0, 5, 10)
# 排期执行率:实际 / 计划
plan = (data.get("plan") or {}).get("planned_posts")
if plan:
rate = ss.safe_div(len(per_post), plan)
comps["schedule_execution"] = ss.scale_to_range(rate, 0.5, 1.0, 0, 10)
# 触达趋势:环比方向
rc = mom.get("reach_change_pct")
if rc is not None:
comps["reach_trend"] = ss.scale_to_range(rc, -20, 20, 0, 10)
score = ss.weighted_score(comps, SCORE_WEIGHTS)
return {
"score_1to10": round(score, 1) if score is not None else None,
"components": comps,
"dimensions_used": list(comps.keys()),
"note": "缺数据的维度已剔除并重新归一化权重",
}
def cmd_score(args):
data = load_input(Path(args.input))
print(json.dumps(score_posts(data), ensure_ascii=False, indent=2))
def build_parser():
p = argparse.ArgumentParser(description="月度社媒复盘的确定性计算")
sub = p.add_subparsers(dest="cmd", required=True)
s = sub.add_parser("score", help="逐帖评分+互动率+环比+加权内部评分")
s.add_argument("--input", required=True, help="标准化月度帖子 JSON 路径")
s.set_defaults(func=cmd_score)
return p
def main(argv=None):
args = build_parser().parse_args(argv)
args.func(args)
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |