skill-content-postmortem · Content Postmortem
Content postmortem: two modes — single-post review (dissecting why one published post hit or flopped across hook, structure, topic, platform fit, timing, and visuals, with 3 actionable fixes); pattern extraction (distilling shared traits of hits from multiple posts into repeatable hit formulas and anti-pattern lists).
Aggregation statistics are script-handled (hit vs. ordinary group comparisons, cross-analysis, tag co-occurrence); the LLM only interprets the patterns. Without data, analysis is qualitative and explicitly labeled as such.
Example invocation: "Review this viral note and extract the pattern."
Full brief
Positioning
skill-content-postmortem is a post-hoc content analyzer: it dissects why a single post succeeded or failed, or distills repeatable hit formulas from multiple posts. Postmortems, not pre-publish reviews.
Core capabilities
- Mode A single-post review: hook analysis (opening type, first-3-seconds appeal score), content structure (information density and pacing, highlights and breaks), topic assessment (heat type, pain-point fit, differentiation), platform fit (format, distribution mechanics), timing (active hours, news windows), visuals/cover; outputs 3 actionable fixes.
- Mode B pattern extraction: scripts handle aggregation (threshold splits, hit vs. ordinary group comparison, two-dimension cross, tag co-occurrence, sample-size warnings); the LLM interprets high-lift combinations into hit formulas (topic + hook + structure) and anti-formulas.
- Hit formula templates: title formula (emotion word + number + suspense/contrast), structure formula (hook → pain resonance → solution → CTA), topic formula (news event × niche × counter-intuitive angle) — each with transfer conditions and application examples.
- Dimension scoring scale: six 1–10 dimensions (hook strength, structure, information density, engagement prompts, visual layout, platform fit); topic and timing are qualitative.
- Profile integration: with a profile, reviews sharpen around account positioning, audience, and platform history; advice adapts to account stage (cold start / growth / monetization).
Workflow
- Confirm mode (single review / pattern extraction)
- Gather context (platform, publish time, metrics; can proceed when missing)
- Mode A: multi-dimension teardown → attribution ranking (content factors vs. luck) → fix prescription
- Mode B: standardized data intake → script aggregation → LLM pattern interpretation → formulas and action list
- Output report, saved to
outputs/
Inputs & outputs
| Input | Notes |
|---|---|
| Post text or link | Required for Mode A — full published post (title + body + tags) |
| Platform | Required for Mode A; per-post for Mode B |
| Metrics | Recommended — views/plays, likes, saves, comments, shares, completion rate |
| Multiple posts' data | Required for Mode B — at least 5 posts with titles, platforms, core metrics |
Output: single-post review (verdict snapshot, multi-dimension teardown, fix prescription) or hit-pattern report (shared hit traits, hit formulas, repeatable action list, data limitations).
Fit
- A post went viral or flopped and you want to know why
- Distilling a repeatable hit methodology after accumulating posts
- Source material for team postmortem meetings
Before you start
- Prepare post text and metrics; without data, analysis is qualitative and labeled as such.
- Pattern extraction needs at least 5 posts; the script warns on small samples.
skill-content-postmortem is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.