What's Wrong: translation QA
Checks translations against their source text, with reports for language teams
The server answered yesterday with a sign-in request for OAuth (HTTP 401), so we could not list its tools.
What's Wrong: translation QA is a freemium MCP server for translation quality. It runs as a hosted remote server, so there is nothing to install locally.
- Pricing
- Freemium
- Platforms
- Web, API, CLI
- Runs
- Remotely
- Access
- Needs sign-in (OAuth)
- Version
- 0.1.0
- Listing updated
Similar MCP servers
Install
Run in a terminal
claude mcp add --transport http whats-wrong https://app.stillneedsahuman.com/mcpAdd to Settings → Connectors → Add custom connector
https://app.stillneedsahuman.com/mcpAdd to ~/.cursor/mcp.json
{
"mcpServers": {
"whats-wrong": {
"url": "https://app.stillneedsahuman.com/mcp"
}
}
}Add to .vscode/mcp.json
{
"servers": {
"whats-wrong": {
"type": "http",
"url": "https://app.stillneedsahuman.com/mcp"
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"whats-wrong": {
"serverUrl": "https://app.stillneedsahuman.com/mcp"
}
}
}Add to Grok Bot desktop app → Connect apps → Add a custom connector
https://app.stillneedsahuman.com/mcpEnter a Name (for example "What's Wrong: translation QA"), choose Type: Remote HTTPS, paste this as the Server URL and choose Add plugin. When we checked, this server asked for OAuth sign-in. If its docs give you an OAuth client ID and secret, enter them in the form's OAuth Client ID and Client Secret fields.
Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"whats-wrong": {
"type": "http",
"url": "https://app.stillneedsahuman.com/mcp"
}
}
}Using several servers? Build one config file with the MCP config generator.
About What's Wrong: translation QA
Translation QA server that lets an AI assistant such as Claude or ChatGPT run a quality check on a translation and return the results. It compares a target text with its source, applies the workspace's profile and glossary rules, and hands back a score, counts, findings and a link to a full report. It does not translate text itself.
The check runs on the same engine as the web app. Assistants can run checks on short texts or uploaded files, read existing reports and findings, list segments, and check the status of a run. Where the workspace permits, they can also mark findings as ignored or fixed, write confirmed fixes into a copy of a file, and draft or change translation profiles, but only after a person has approved the change. Reports are never signed off through the connector.
It suits localization managers, translators, reviewers and agencies who want quality checks inside a chat workflow rather than switching to a separate tool. Access is tied to the user's own role and permissions, and the connector is enabled per workspace.
Tools this server provides (17)
Taken from the project documentation; the running server may differ.
- check_text
- Checks a source text and its translation with the workspace's profile.
- list_profiles
- Lists the profiles the workspace can run.
- describe_profile
- Describes what one profile checks.
- list_reports
- Lists reports, newest first, with result, score and link.
- get_report
- Returns one report with result, score, counts and findings.
- list_findings
- Lists every finding of a report with filters.
- get_segment
- Returns one segment with its source, target and findings.
- get_run_status
- Returns a run's state and progress.
- get_usage
- Returns the month's AI word usage and reset date.
- request_upload
- Creates a one-time upload link for a file.
- get_upload
- Returns the state of an upload link.
- check_file
- Checks one file from an upload link, Dropbox path or inline content.
- check_files
- Checks up to five files in one call.
- list_dropbox
- Lists files in watched Dropbox folders.
- apply_fixes
- Writes confirmed fixes into a copy of a file with a change log.
- recheck
- Reruns a fixed file and reports resolved, remaining and new findings.
- set_finding_status
- Marks findings as ignored, fixed or open with a reason.
What it does
- Source and target checking
- Checks a translation against its source text, optionally with a chosen profile.
- Report retrieval
- Lists reports and findings with filters for profile, state, result and date.
- Segment inspection
- Shows the source, target and every finding for a single segment.
- Confirmed fix copies
- Writes confirmed fixes into a copy of the file, never the original, with a change log.
- Recheck after fixing
- Reruns a fixed file and reports what was resolved, what remains and what is new.
- Permission-scoped access
- Assistant actions follow the user's role and the permissions they grant on connection.
Good for
- Run a translation check from Claude or ChatGPT without opening the web app
- Review findings from a previous report and see which ones a second opinion agreed with
- Mark reviewed glossary findings as fixed with a stated reason
- Check the month's remaining AI word allowance before queuing more files
- Build a translation profile from a client's folder of style guides and glossaries
Pricing
Freemium
Free plan with a monthly AI word allowance; paid Solo, Team and Business plans; rule checks are not metered
Prices change. Check stillneedsahuman.com before you buy.
What's Wrong: translation QA FAQ
Does the connector translate text?
No. It checks an existing translation against its source.
How do I connect What's Wrong: translation QA?
Add a custom connector in Claude or ChatGPT using the server URL https://app.stillneedsahuman.com/mcp and sign in with OAuth when prompted.
Can the assistant sign off a report?
No. Reports are signed off by a person in the app, and the connector never signs a report off.
Is What's Wrong: translation QA free?
What's Wrong: translation QA has a free plan, with paid plans for more usage or features.
How do I add What's Wrong: translation QA to Claude Code?
Run this command in your terminal: claude mcp add --transport http whats-wrong https://app.stillneedsahuman.com/mcp
Where this comes from
- Description: written from the project's website and its MCP Registry entry.
- Status: checked automatically by AIPoweredBot on 21 November 2026.
- Install steps: from the official MCP Registry entry.