tune-ai-search-portal — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited tune-ai-search-portal (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
| Boost Type | Parameter | Range | Impact |
|---|---|---|---|
| Field boost | title weight | 2.0-5.0 | Title matches ranked higher |
| Freshness | date decay | linear/exponential | Recent content preferred |
| Magnitude | popularity score | 1.0-3.0 | Popular docs ranked higher |
| Tag boost | category match | 1.5-3.0 | Category-filtered results boosted |
| Distance | geo proximity | 1.0-2.0 | Nearby results preferred |
Scoring profile template:
{
"name": "production-scoring",
"text": { "weights": { "title": 3.0, "content": 1.0 } },
"functions": [
{ "type": "freshness", "fieldName": "date", "boost": 2.0, "interpolation": "linear", "freshness": { "boostingDuration": "P30D" } },
{ "type": "magnitude", "fieldName": "viewCount", "boost": 1.5, "magnitude": { "boostingRangeStart": 0, "boostingRangeEnd": 1000 } }
]
}| Configuration | Keyword | Vector | Semantic | Best For |
|---|---|---|---|---|
| Keyword-heavy | 0.6 | 0.2 | 0.2 | Exact term matching (codes, IDs) |
| Balanced | 0.33 | 0.33 | 0.34 | General-purpose search |
| Vector-heavy | 0.2 | 0.5 | 0.3 | Conceptual/semantic queries |
| Semantic-first | 0.2 | 0.2 | 0.6 | Natural language questions |
Diagnostic: Compare NDCG@10 across configurations to find optimal weights.
| Parameter | Default | Range | Impact |
|---|---|---|---|
| Replicas | 1 | 1-12 | More = higher read throughput |
| Partitions | 1 | 1-12 | More = larger index capacity |
| Analyzer | standard | standard/custom/language | Language-specific = better recall |
| Synonym maps | None | 0-5 maps | Improves zero-result rate |
Scaling rule: queries/second ÷ 15 = required replicas (approximate).
| Model | Dimensions | Cost/1M tokens | Quality | Latency |
|---|---|---|---|---|
| text-embedding-3-large | 3072 | $0.13 | Highest | ~50ms |
| text-embedding-3-small | 1536 | $0.02 | Good | ~30ms |
| text-embedding-ada-002 | 1536 | $0.10 | Legacy | ~40ms |
Recommendation: Start with text-embedding-3-large for quality, switch to small if cost is critical and quality delta < 5%.
| Component | Cost Driver | Optimization |
|---|---|---|
| AI Search | Replicas × partitions × tier | Right-size: Standard S1 with 1 replica for dev |
| Embeddings | Documents × reindex frequency | Incremental indexing, not full reindex |
| Storage | Document volume | Archive old versions, compress |
| Queries | QPS × complexity | Cache frequent queries in Redis |
Monthly cost estimate (100K documents, 10K queries/day):
After tuning, compare before/after:
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.