# Data Quality Gate - deterministic post-scrape cleaner + verdict

> Post-scrape data cleaner, no LLM: repairs mojibake, HTML, invisible chars. Plus a verdict.

- name: io.github.aidatatools-dev/data-quality-gate
- version: 0.3.0
- connection class: R0 (autonomous)
- trust: 77/100 flags: no-repository
- quality: 80/100 (good)
- owner: not claimed
- descriptor: https://api.protogrid.dev/v1/servers/io.github.aidatatools-dev%2Fdata-quality-gate
- tools: https://api.protogrid.dev/v1/servers/io.github.aidatatools-dev%2Fdata-quality-gate/tools

**An agent can connect right now, no human step.** Remote endpoint, no authentication, reachable on the last probe.

## Connection (mcpServers)

```json
{
  "mcpServers": {
    "data-quality-gate": {
      "type": "http",
      "url": "https://www.aidatatools.dev/api/mcp_server"
    }
  }
}
```


## Trust

- hygiene: 80
- liveness: 100
- freshness: 85
- provenance: 45
- drivers: +website +reachable +uptime-100% ~updated-69d-ago -no-repository
- Derived from observable signals (official registry feed and our own credential-free probes); no code audit performed.

## Quality 80/100 (good)

### protocol: 88 (weight 25)

- pass `protocol.modern`: supports 2026-07-28 (server/discover)
- pass `protocol.stateless`: answers without a session
- pass `protocol.transport`: streamable HTTP
- warn `protocol.list_ttl`: tools/list declares ttlMs 0, so clients re-fetch it on every use

### authorization: n/a (weight 15)

- n/a `auth.prm`, `auth.as_metadata`, `auth.cimd`: no remote uses OAuth
- n/a `auth.secret_in_url`: no templated URL

### tool hygiene: 64 (weight 30)

- pass `tools.descriptions`: every tool has a description
- warn `tools.description_length`: 2 tools have descriptions over 1,024 characters, which crowds the context
- pass `tools.schemas`: every tool has an object input schema
- fail `tools.annotations`: no tool declares readOnlyHint or destructiveHint, so clients cannot tell safe calls from risky ones
- fail `tools.directory_hints`: no tool declares all four of readOnlyHint, destructiveHint, idempotentHint and openWorldHint; missing: check_dataset_quality (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); clean_scraped_data_audited (readOnlyHint, destructiveHint, idempotentHint, openWorldHint); clean_scraped_data (readOnlyHint, destructiveHint, idempotentHint, openWorldHint)
- pass `tools.token_cost`: about 1,927 tokens to load every tool
- pass `tools.api_dump`: tools are not a one-to-one API dump

### stability: 100 (weight 15)

- pass `stability.changes`: no tool changes in 30 days
- pass `stability.rug_pull`: no tool changed its meaning under the same name

### dependencies: n/a (weight 15)

- n/a `deps.known_vulns`, `deps.mcp_sdk_version`, `deps.resolvable`: no npm or PyPI package

- tools: 3, about 1,927 tokens to load them all
- tool set hash: 2690ec8bfd9bfd7e8528a43ec642c0fb
- badge: [![protogrid quality](https://protogrid.dev/badge/io.github.aidatatools-dev/data-quality-gate.svg)](https://protogrid.dev/servers/io.github.aidatatools-dev/data-quality-gate)
- Checks run on what our credential-free, read-only probes observe; tools are never called and no code audit is performed.

### Tool changes

_No tool definition changes recorded._

## Remotes

- https://www.aidatatools.dev/api/mcp_server (streamable-http, auth none, reachable true, uptime30d 1)

## Packages

_none_

## Tools (3)

- `check_dataset_quality`: Call this before using any dataset. Returns a deterministic quality verdict (RELIABLE / USABLE_WITH_CLEANING / UNRELIABLE) with exact facts: completeness, nulls, type consistency, impossible values, duplicates, outliers, and (on financial/trading data) cross-source price divergence. 100% deterministic, no LLM. Free -- this MCP endpoint runs the engine directly; POST /api (plain REST, same engine) is x402-gated at $0.01/call instead. Input: rawJson (a JSON array of row objects, or a single object); datasetId is accepted but not resolvable on this deployment -- pass rawJson instead.
- `clean_scraped_data`: PAID TIER, CALLED FROM A FREE ENDPOINT -- READ THIS FIRST: this tool does NOT return your cleaned data, and its response has NO `data` key. It returns {"status": "payment_required", "what_we_found_in_your_data": {...}, "get_the_repaired_data": {...}}. Do not treat its output as cleaned data and do not write it back into your dataset.
WHAT YOU GET HERE, FREE: an exact inventory of the defects in YOUR data -- which repair rules would fire, how many values each would change, and which cases cannot be repaired safely and need your decision.
HOW TO GET THE REPAIRED DATA: make the one call named in `get_the_repaired_data` -- POST https://www.aidatatools.dev/api/clean, $0.04 via x402, no account, no API key, no signup. That response body IS the cleaned dataset, in the shape you posted it.
WHY THE SPLIT: detection is free on this endpoint and always has been (check_dataset_quality reports the same defects). The repaired artifact is the paid product, because it is re-bought on every extraction run rather than cached like a verdict.
WHAT THE PAID CALL DOES: removes leftover HTML tags and entities, decodes mojibake ('Café' -> 'Café'), strips invisible characters (zero-width, BOM, soft hyphen), normalises non-breaking spaces and trims values -- across nested objects and arrays too. 100% deterministic, no LLM: the same input always yields byte-identical output, and cleaning twice equals cleaning once. It repairs how data was ENCODED, never what it SAYS: masked placeholders ('N/A', 'None'), near-duplicate rows and failed extractions ('access denied', 'captcha', which mean that record must be re-scraped) are reported with a proposal, never silently deleted or rewritten. The full boundary -- 7 rules applied automatically, 5 needing an explicit opt-in, 8 only ever reported -- is at GET https://www.aidatatools.dev/api/clean.
- `clean_scraped_data_audited`: PAID TIER, CALLED FROM A FREE ENDPOINT -- READ THIS FIRST: this tool does NOT return your cleaned data, and its response has NO `data` key. It returns {"status": "payment_required", "what_we_found_in_your_data": {...}, "get_the_repaired_data": {...}}. Do not treat its output as cleaned data and do not write it back into your dataset.
WHAT YOU GET HERE, FREE: an exact inventory of the defects in YOUR data -- which repair rules would fire, how many values each would change, and which cases cannot be repaired safely and need your decision.
HOW TO GET THE REPAIRED DATA: make the one call named in `get_the_repaired_data` -- POST https://www.aidatatools.dev/api/clean/audit, $0.12 via x402, no account, no API key, no signup. That response body IS the cleaned dataset, in the shape you posted it.
WHY THE SPLIT: detection is free on this endpoint and always has been (check_dataset_quality reports the same defects). The repaired artifact is the paid product, because it is re-bought on every extraction run rather than cached like a verdict.
WHAT THE PAID CALL DOES: the same repair as clean_scraped_data, plus a complete audit trail: every transformation with its path, rule, before and after value, a replay_id, and input/output SHA-256. The ledger is a full inverse patch -- applying it in reverse reconstructs your original input byte for byte. Use it when you must be able to PROVE later what changed and why.