validate-content

---

name: validate-content description: Universal content validator. Orchestrates the check chain — check-facts → check-generic → check-sources → type-specific checks → detect-ai → check-generic (2nd pass). Works for posts, comments, profiles, PR pieces, and event promo. disable-model-invocation: true argument-hint: [path-to-content-file] [--type post|comment|profile|pr|event-promo] allowed-tools: Read, Grep, Glob, Skill(check-facts *), Skill(check-generic *), Skill(check-sources *), Skill(detect-ai *)

Universal Content Validator

Full re-validation of content against ALL canons of the LinkedIn Content Machine via a universal validation chain.

When to run:

• Automatically after creating a post via /create-post
• After ANY content edits
• Final check before publication
• For comments, PR comments, profile text, promo

Input

Content file: $ARGUMENTS

Determine --type from arguments or from the file contents:

• Has frontmatter with ## Post → type = post
• Has ## Comment or comment context → type = comment
• Has ## Headline / ## About → type = profile
• Has ## Commentary → type = pr
• Has ## Promo → type = event-promo
• Not detectable → ask the user

---

PHASE 1: Universal Validation Chain

Invoke in sequence (each runs in isolation):

1. /check-facts $ARGUMENTS — facts and logical integrity
2. /check-generic $ARGUMENTS — generic language and repetition (FIRST PASS)
3. /check-sources $ARGUMENTS — transparent source map

Collect results from all three.

---

PHASE 2: Type-Specific Checks

Run checks specific to the content type.

If type = post

A. Characters and length

Count characters in the ## Post section (text only, no frontmatter/Brief/Hook Alternatives):

Range	Verdict	Comment
1,200–1,800	OPTIMAL	Sweet spot: +47% engagement
500–1,200	OK	For a bold statement / hot take
1,800–2,500	ACCEPTABLE	Only with a strong hook + story-driven
2,500–3,000	LONG	Only for exceptional content
>3,000	ERROR	LinkedIn limit exceeded
<500	SHORT	Confirm this was intentional

B. Format and rotation

1. Extract format: from frontmatter
2. Read the author's last 5 posts in Posted/[Author]/
3. If the format matches 2+ of the last 3 → WARNING with an alternative from Format-Performance-Data.md

C. Cluster and pillar

1. Extract cluster: from frontmatter
2. Compare with DIP Expertise Clusters
3. If outside the DIP clusters → WARNING: Profile-Content Alignment
4. If 5+ posts in a row in the same cluster → WARNING: monotony

D. Hook

1. First 1–2 lines of ## Post
2. If >140 characters → WARNING: it'll be cut by "See more"
3. Compare with hooks from the last 10 posts → if the structure repeats → WARNING

E. Anti-Patterns

Load Knowledge-Base/03-Viral-Trends/Anti-Patterns.md and check:

• No engagement bait?
• No "bro-etry"?
• No "we/our" in a personal post?
• No link in the first 300 characters?
• Not posting within 12h of the previous one?
• No generic motivational quotes?

F. LinkedIn algorithm

Load Algorithm-Intelligence.md:

• Structure is readable? Short paragraphs, whitespace
• CTA is specific?
• Profile-Content Alignment?

G. Mobile readability

• Paragraphs ≤3 lines?
• Blank lines between blocks?
• No walls of text?

If type = comment

• Minimum 15 words?
• First person (I/my)?
• Not generic ("Great post!", "Love this!")?
• Not selling/pitching?
• Doesn't duplicate the original post?
• No "not X but Y" constructions?
• No "I've seen..." without specifics?
• Source grounding passed? (verified fact / grounded inference / question)

If type = profile

• Headline ≤120 characters?
• About section: hook in the first 2 lines?
• No generic descriptions?
• Specifics: numbers, results, achievements?
• Tone of voice = DIP?
• Profile-Content Alignment with clusters?

If type = pr

• ≤150 words (PR requirement)?
• Quotable sentence (pullable into a headline)?
• Not marketing-speak?
• Tone = the author's DIP?
• Specifics (numbers, facts, experience)?

If type = event-promo

• ICP fit: the topic is relevant to the target audience?
• No generic event descriptions?
• Concrete value proposition for the attendee?
• CTA is clear?

---

PHASE 3: AI Detection & Humanization

Invoke:

4. /detect-ai $ARGUMENTS — finds AI patterns and suggests fixes

---

PHASE 4: Second Pass — Check Generic #2

After detect-ai, the humanizer may have introduced new generic phrasing. Final pass:

5. /check-generic $ARGUMENTS --second-pass — verify the humanizer didn't introduce new generic content

---

PHASE 5: Summary report

## Full validation result

### File: [path]
### Content type: [post / comment / profile / pr / event-promo]
### Author: [owner]
### Check date: [today]

---

### Summary

| Check | Status | Detail |
|-------|--------|--------|
| Facts (check-facts) | OK / ATTENTION / CRITICAL | [brief] |
| Generic Pass 1 (check-generic) | CLEAN / NOTES / REWORK | [brief] |
| Sources (check-sources) | OK / ATTENTION | [N] direct, [N] compilations, [N] no source |
| Type-specific checks | OK / NOTES | [brief] |
| AI Detection (detect-ai) | HUMAN / LOW AI / MEDIUM AI / HIGH AI | Score: [N] |
| Generic Pass 2 (check-generic) | CLEAN / NEW ISSUES | [brief] |

---

### Detail: Fact-check
[check-facts result]

### Detail: Generic Pass 1
[check-generic first-pass result]

### Detail: Source Map
[check-sources result]

### Detail: Type-Specific Checks
[per-type results]

### Detail: AI Detection
[detect-ai result]

### Detail: Generic Pass 2
[check-generic second-pass result]

---

### Final recommendations (prioritized)

**Critical (must fix):**
1. [...]

**Important (recommended fixes):**
1. [...]

**Optional (nice to have):**
1. [...]

After edits

The user can re-run /validate-content [file]. All checks will rerun against the updated content.