Content Reviewer

Content Reviewer

Systematically review influencer content submissions against brand guidelines, campaign objectives, and legal requirements, then issue a gate decision before the content goes live.

When This Must Trigger

Run this before any influencer content goes live, even if the user doesn't use review terminology:

• User asks "can we post this", "is this submission okay", or "does this meet our guidelines"
• A creator delivers a draft (post, video, image, caption) against a brief from brief-generator
• User suspects a disclosure, claim-substantiation, or brand-safety problem in submitted content
• Before content-amplifier puts paid or owned reach behind a creator post
• Compliance re-check when FTC/platform disclosure rules or campaign claims change mid-flight

Quick Start

Shortest invocation:

Review this influencer content submission: [content description/link]

Common scenario — gate a submission against a campaign brief:

Check if this TikTok post meets our brand guidelines for [campaign], then give me an approve/revise/reject decision and feedback I can send the creator

Skill Contract

• Reads: the content submission (link, description, caption, media notes), the campaign brief and key messages, brand guidelines, disclosure/FTC requirements, platform technical specs, prior review artifacts in memory/audits/influencer/, and the auditor-runbook handoff/cap schema (content-reviewer is the C³ ART gate consumer).
• Writes: a gated review artifact at memory/audits/influencer/YYYY-MM-DD-<topic>.md with class: auditor-output and the auditor-runbook handoff schema (status / key_findings / evidence_summary / recommended_next_skill / cap_applied / raw_overall_score / final_overall_score), computed from C³ ART scoring — so the PostToolUse Artifact Gate validates the verdict — plus the human-facing feedback message.
• Promotes: durable facts (final approval status, recurring compliance gaps, creator-specific notes) to memory/hot-cache.md.
• Done when:
  1. Every must-pass category (brand alignment, message accuracy, compliance, technical specs) has an explicit pass/fail with evidence.
  2. A single decision is recorded — APPROVED, APPROVED WITH MINOR CHANGES, REVISIONS REQUIRED, or REJECTED.
  3. A constructive feedback message for the creator exists (when changes are needed).
  4. The gated artifact follows the auditor-runbook schema: raw_overall_score = the ART score; a T1/T2 veto → status: BLOCKED (Reject) with cap_applied: false, raw_overall_score retained, and no final_overall_score; a non-veto pass → status: DONE/DONE_WITH_CONCERNS with final_overall_score; a non-veto REVISIONS decision → status: NEEDS_INPUT, still with final_overall_score (cap_applied: false) — BLOCKED is the only branch that omits the final score. Map the decision to the status enum (APPROVED→DONE, MINOR→DONE_WITH_CONCERNS, REVISIONS→NEEDS_INPUT, REJECTED→BLOCKED).
• Primary next skill: contract-helper

Handoff Summary

Emit the standard shape from skill-contract.md §Handoff Summary Format.

Specifically, emit the auditor-class handoff from auditor-runbook.md §1: status (DONE / DONE_WITH_CONCERNS / BLOCKED / NEEDS_INPUT), objective, target, key_findings, evidence_summary, recommended_next_skill, plus the auditor fields cap_applied, raw_overall_score (the ART score, floor-rounded, before cap), and final_overall_score (omitted only when BLOCKED).

Data Sources

This family needs no live integrations (Tier 1). The skill works end to end from inputs you provide: paste the content link or description, the brief, brand guidelines, and platform specs, and it produces the review.

Optional connectors can speed up evidence gathering where available:

• ~~social platform analytics — pull native post specs, engagement signals, and platform disclosure flags (e.g. "Paid partnership" label).
• ~~influencer database — load the creator's history, prior submissions, and past compliance flags.
• ~~CRM — link the review back to the deal, campaign owner, and approval chain.
• ~~compliance reference — current FTC / platform disclosure rules for the relevant market.

With no connector configured, ask the user for the missing inputs and proceed. See CONNECTORS.md for the keyless data recipes.

Instructions

Before scoring, Read ../../references/auditor-runbook.md and ../../references/c3/art-content-benchmark.md. The runbook is the framework-agnostic SSOT (§1 handoff schema, §2 cap method, §4 Artifact Gate, §5 veto-ID translation); the ART benchmark owns the Content dimensions and the T1/T2 veto definitions this gate enforces.

Run the review steps in order. Each step has a fill-in template in references/review-templates.md.

1. Establish review criteria — set campaign, creator, platform, content type, and the weighted category table (brand alignment, message accuracy, compliance must-pass; quality 80%+; technical must-pass). Template Step 1.
2. Brand alignment review — visual brand check, brand-safety check, value alignment; score X/10. Template Step 2.
3. Message accuracy review — key messages, talking points, prohibited claims, CTA; score X/10. Template Step 3.
4. Compliance review — disclosure, platform rules, legal/regulatory, required elements; score X/10 (no partial pass). This is the C³ ART veto gate — see below. Template Step 4.
5. Quality assessment — production, effectiveness, platform optimization, creative; score X/10. Template Step 5. Run the SOFT AI-slop / persona / per-platform aids in references/quality-review-aids.md.
6. Technical specifications check — platform specs and caption requirements; pass/fail. Template Step 6.
7. Generate final review — submission details, score table, single decision, feedback pointer. Template Step 7.
8. Create feedback message — constructive, positives-first message for the creator when changes are needed. Template Step 8.

C³ ART veto enforcement

Step 4's compliance gate is the C³ Content review (ART). Two ART items are veto-level — a failure forces the overall decision to Reject (never "revise"). Record the veto ID (T1/T2) in the gated artifact and handoff YAML only; the decision and creator feedback the user sees carry the plain-language translation per auditor-runbook §5:

• T1 · FTC Disclosure (Disclosure Check + FTC compliance rows) — missing or inadequate disclosure on sponsored content → Reject. Regulatory basis: FTC 16 CFR §255 and the 2024 Trade Regulation Rule (16 CFR Part 465). Not legal advice.
• T2 · Claim Integrity (Claims substantiation) — false or unsubstantiated claims → Reject.

ART veto-ID translation rows (use alongside the runbook's shared rows — these are the C³ ART meanings, never CORE-EEAT/CITE/ROAS):

Worked Example

User: "Review this TikTok video submission for our protein powder campaign"

A submission with strong visuals (Brand 9/10, Quality 9/10) but no visible #ad disclosure scores Compliance 5/10 and triggers the FTC-disclosure veto (T1 — recorded in the artifact/handoff only). The user-facing decision is ❌ REJECTED — missing the required ad disclosure, not a revise: the creator must add #ad in caption and/or verbal disclosure in the first 3 seconds, then resubmit. Non-veto gaps (missing "20g protein" key message, missing promo code "FIONA20") are listed as required-on-resubmission, and the feedback message leads with what worked (authentic workout integration, engaging hook). Full output in references/review-templates.md § Worked Example.

Validation Checkpoints

Input Validation

• Content submission identified (link, file, caption text, or description)
• Campaign brief and key messages available (else NEEDS_INPUT — the review has no criteria without them)
• Platform and content type confirmed (technical specs and disclosure norms differ per platform)
• Disclosure context known: paid, sponsored, gifted, or organic

Output Validation

• Every must-pass category (brand alignment, message accuracy, compliance, technical specs) has an explicit pass/fail with evidence
• ART vetoes T1 (FTC disclosure) and T2 (claim integrity) checked; any veto forces REJECTED, never "revise"
• Exactly one decision recorded (APPROVED / APPROVED WITH MINOR CHANGES / REVISIONS REQUIRED / REJECTED) and mapped to the status enum
• cap_applied, raw_overall_score, final_overall_score set per the runbook (final omitted only when BLOCKED — NEEDS_INPUT still carries it)
• Decision and creator feedback use plain language; veto IDs and internal field names appear only in the gated artifact/handoff YAML

Reference Materials

• references/review-templates.md — fill-in templates for every step, the full worked example, the quick review checklist, and review tips.
• references/quality-review-aids.md — wires the AI-slop checklist (humanizer-slop.md) as a SOFT, non-veto ART quality penalty (ART vetoes stay T1/T2 only), the multi-persona method (expert-panel.md), and per-platform format/disclosure norms (platforms/).
• skill-contract.md — shared contract and Handoff Summary format.
• state-model.md — memory tiers and save-path conventions.
• CONNECTORS.md — keyless data recipes for the ~~ connector placeholders.
• C³ scoring: c3-benchmark.md and c3/art-content-benchmark.md — the ART Content rubric this review gates on, incl. the T1 (FTC disclosure) and T2 (claim integrity) veto items.
• offer-claims-registry — the claim-substantiation record consulted for the T2 check.
• creator-registry — the creator's disclosure/compliance history; log dated verdicts back to it.
• brief-generator — the brief whose criteria this review checks against.
• contract-helper — bake content guidelines and approval terms into the agreement.
• content-amplifier — amplify content once approved.
• performance-analyzer — track how approved content performs.

Next Best Skill

Primary: contract-helper — once content is approved (or once review feedback exposes a gap the agreement should cover), fold guidelines, usage rights, and approval terms into the partnership contract.

Alternates (same Activate family):

• content-amplifier — when content is approved and ready to scale through paid or owned channels.
• performance-analyzer — when the piece is live and you want to measure outcomes.

Termination note (visited-set): if a recommended skill has already been invoked this session, stop and report the chain as complete rather than re-running it. Cap any handoff chain at max-depth 3.