菜单
Analyze post performance. Compares with baseline, detects patterns, evaluates experiments, analyzes comments, updates ICP. Lifecycle-aware.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | analyze |
| description | Analyze post performance. Compares with baseline, detects patterns, evaluates experiments, analyzes comments, updates ICP. Lifecycle-aware. |
| user-invocable | true |
| allowed-tools | ["Read","Write","Edit","Agent","Glob","Grep"] |
Analyzes post performance, detects patterns, and sharpens the ICP. Lifecycle-aware: distinguishes Active/Cooling/Archived posts.
IMPORTANT: Delegate the work to the post-analyzer agent. Do NOTHING yourself — start the agent with the Agent tool and pass the post URN.
/analyze <urn> # Analyze single post
/analyze # All posts that need analysis
/analyze patterns # Pattern refresh only (archived posts)
/analyze icp # ICP sharpening only
| Lifecycle | Analysis Behavior |
|---|---|
| Active (0-7 days) | In data/posts/ — analysis possible, but with caveat "metrics still moving" |
| Cooling (7-14 days) | In data/posts/ — good data basis, snapshot 2-3 available |
| Archived (14+ days) | In data/posts/archive/ — final metrics, best basis for pattern detection |
Pattern refresh uses only archived posts (data/posts/archive/).
post-analyzer agentPost Analysis: "Most SMBs underestimate..."
URN: urn:li:activity:7435982583777169408
Lifecycle: Cooling (Day 9)
Performance (Snapshot 2):
Reactions: 89 (Baseline: 45) → +98%
Comments: 12 (Baseline: 5) → +140%
Impressions: 3,200 (Baseline: 1,800) → +78%
Engagement Rate: 3.2% (Baseline: 2.8%) → +14%
Content Properties:
Hook: Surprising Fact | Format: Text | CTA: Question
Day: Tuesday | Hour: 8 | Length: Medium (780 chars)
Pattern Match:
"Surprising Fact" hook → above average
"Tuesday" posting → confirms pattern
"Personal Reference" → +40% (Low Confidence, n=4)
ICP Insight:
Top Engagers: CTOs (30%), Developers (40%), PMs (20%)
Next snapshot in 5 days (Day 14).
Only on archived posts (data/posts/archive/):
Complete reference for the linkedin-cli tool. Load this skill when you need to call the CLI via Bash to fetch LinkedIn data, manage connections, or send messages.
Morning Check / Session Entry Point. Delta pipeline: Stage 1 COLLECT → Stage 2 DETECT (+ Feed parallel). On-the-fly engagement analysis, no stored contacts.
Quick status check without API calls. Shows current state from data files + session info.
Complete data schema for LinkedIn tracking. Markdown file-per-record with YAML frontmatter. Naming conventions, schemas for all 11 record types, Glob/Read/Write/Edit access patterns.
Generate personalized outreach message for a LinkedIn contact.
Deep onboarding for LinkedIn Commander. Interview + historical post analysis + competitor deep dives + system generation. The system starts "warmed up", not empty.