From arb
Classifies page or content type for SCB, resolves profile-specific weights and extension items, and outputs normalized scoring manifest for downstream SCB skills and reports.
npx claudepluginhub huifer/adsense-readiness-benchmark --plugin arbThis skill uses the workspace's default tool permissions.
Profile-aware routing layer for SCB v2.0. Detects the best-fit content profile, applies cross-profile rules, and emits a strict manifest contract used by scoring and remediation skills.
Suggests manual /compact at logical task boundaries in long Claude Code sessions and multi-phase tasks to avoid arbitrary auto-compaction losses.
Share bugs, ideas, or general feedback.
Profile-aware routing layer for SCB v2.0. Detects the best-fit content profile, applies cross-profile rules, and emits a strict manifest contract used by scoring and remediation skills.
Route content into the correct SCB profile logic before scoring:
Input: URL or source path + optional profile hint Output: Profile manifest in Markdown + JSON Optional Scripts: Content classifier, schema parser, heading/entity extractor Time: 5-12 minutes per URL/page
| Field | Required | Default | Notes |
|---|---|---|---|
| url_or_path | Yes | - | Target page URL or local source path |
| score_mode | No | Core 80 + Profile | Core 80 / Core 80 + Profile / Full 106 |
| ymyl | No | auto | true / false / auto |
| language_scope | No | single | single / multilingual |
| profile_hint | No | none | long-form/news/review/how-to/ymyl/tool/comparison/ugc |
| query_context | No | none | target query or keyword cluster for better routing |
This router does not assign final item scores. It defines scope and weighting contract for downstream scoring.
Mandatory output fields:
Gather structural and semantic signals:
Map to one of 8 SCB profiles:
If profile_hint is present, treat it as prior evidence, not an override.
Apply cross-profile and conditional rules:
Default rule:
Output must include an explanation for each merged pillar weight.
Compute final item set by score mode:
Mark non-applicable items explicitly in na_items; do not silently drop.
Export normalized JSON and summary markdown for downstream skills.
digraph scb_profile_router {
"Collect Inputs" [shape=box];
"Extract Signals" [shape=box];
"Classify Primary Profile" [shape=box];
"Detect Secondary/Triggers" [shape=box];
"Resolve Weights" [shape=box];
"Build Item Scope" [shape=box];
"Emit Manifest" [shape=doublecircle];
"Collect Inputs" -> "Extract Signals";
"Extract Signals" -> "Classify Primary Profile";
"Classify Primary Profile" -> "Detect Secondary/Triggers";
"Detect Secondary/Triggers" -> "Resolve Weights";
"Resolve Weights" -> "Build Item Scope";
"Build Item Scope" -> "Emit Manifest";
}
Always return these as watchlist candidates:
# SCB Profile Manifest
- URL: https://example.com/article
- Score Mode: Core 80 + Profile
- Selected Profile: Comparison and Alternative (0.89)
- Secondary Profiles: Product Review (0.63)
- YMYL: false
- Multilingual: true
## Resolved Weights
- IQ: 20%
- EA: 25%
- IA: 20%
- FC: 15%
- UX: 15%
- GE: 10%
## Item Scope
- Covered Items: 88
- Selected Profile Items: [IQ15, IQ06, EA16, EA14, FC04, FC15, GE05, GE11]
- Triggered Extension Items: [IA17, GE16]
- Veto Watchlist: [IQ12, IA02, IA10, FC14]
{
"meta": {
"url_or_path": "https://example.com/article",
"score_mode": "Core 80 + Profile",
"router_version": "1.0",
"timestamp": "2026-05-05T00:00:00Z"
},
"routing": {
"selected_profile": "comparison",
"routing_confidence": 0.89,
"secondary_profiles": [
{"profile": "review", "confidence": 0.63}
],
"ymyl": false,
"language_scope": "multilingual"
},
"scope": {
"covered_items": ["IQ01", "..."],
"missing_items": [],
"na_items": [],
"selected_profile_items": ["IQ15", "EA14", "GE05"],
"triggered_extension_items": ["IA17", "GE16"],
"veto_watchlist": ["IQ12", "IA02", "IA10", "FC14"]
},
"weights": {
"IQ": 0.20,
"EA": 0.25,
"IA": 0.20,
"FC": 0.15,
"UX": 0.15,
"GE": 0.10,
"merge_explanation": [
"EA took stricter weight from comparison profile"
]
}
}
scb-profile-router feeds into:
Recommended chain:
Problem: best tools pages are routed as single-product reviews Fix: if 2+ named products and comparison table exists, elevate comparison profile
Problem: educational guides with few steps are routed as how-to Fix: require explicit task completion sequence for how-to
Problem: finance/legal pages routed as general long-form Fix: apply stricter keyword and entity risk lexicon before final classification
Related Skills: