// SCAN REPORT

Localization

46 tools. 26 can modify or destroy data without limits.

7 destructive tools with no built-in limits. Policy required.

Last updated: 5 Jul 2026

26 can modify or destroy data

20 read-only

46 tools total

Verified server · catalogue entry verified 05/07/2026 · full schemas captured for 44 of 46 tools

How to control Localization ↓

TOOL MAP

What Localization exposes to your agents

Read (20) Write / Execute (19) Destructive / Financial (7)

Running more than one server? Check your whole stack →

CONTEXT-WINDOW COST

What Localization costs in tokens

10,053 tokens of tool definitions, loaded on every request

5.0% of a 200k context window

531 heaviest tool: bulk_import

Full per-tool breakdown → Model it in the token calculator →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Localization tools

Destructive · High delete_locale The tool irreversibly deletes translation data at scale (all values for an entire locale across all namespaces). Destructive · High reset_namespace_quality The tool irreversibly clears (destroys) all quality scores across an entire namespace. Destructive · High delete_namespace This tool permanently removes an entire namespace and all associated translation keys and values without possibility of recovery. Destructive · High delete_translation Although the deletion is scoped to a sandbox and marked as 'soft delete', the tool permanently removes translation data that cannot be recovered without external restoration.

26 of Localization's 46 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Localization

PolicyLayer is an MCP gateway — it sits between your AI agents and Localization, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "delete_locale": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "bulk_ai_translate": {
    "limits": [
      {
        "counter": "bulk_ai_translate_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "ai_quality_check": {
    "limits": [
      {
        "counter": "ai_quality_check_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Localization — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON LOCALIZATION →

Instant setup, no code required.

FULL CATALOGUE

All 46 Localization tools

DESTRUCTIVE 7 tools

Destructive delete_locale Remove a locale from a project. This deletes ALL translation values for that locale across all namespaces. Thi Destructive delete_namespace Delete a namespace from a project. This cascades to ALL keys and values in the namespace. This is a destructiv Destructive delete_translation Delete one or more translation keys from the sandbox (soft delete). Keys remain in production until you promot Destructive delete_webhook Delete a webhook configuration. This permanently removes the webhook and stops all future deliveries. Destructive reset_namespace_quality Clear all quality scores in a namespace to trigger fresh re-evaluation. The quality worker will re-assess all Destructive reset_namespace_translations Delete sandbox translations and trigger re-translation. Scope: namespace (no params), single locale ({ locale Destructive reset_sandbox Discard all sandbox changes and re-copy from current production state. This destroys all pending sandbox edits

WRITE 19 tools

Write bulk_ai_translate Alternative bulk translation: translate multiple English texts to all project locales using backend AI (Gemini Write bulk_check_quality Run AI quality check on multiple keys in a namespace at once. When keys is omitted, checks ALL keys in the nam Write bulk_import Before writing to a project for the first time in a session, call assess_integration_state to understand clien Write bulk_set_locale Before writing to a project for the first time in a session, call assess_integration_state to understand clien Write bulk_translate_and_save ALWAYS call assess_integration_state at the start of a new session before writing. Alternative all-in-one: tra Write check_entry_quality Run AI quality check on all locales of a specific translation key in SANDBOX and PERSIST the results to sandbo Write create_locale Add a locale to a project. The primary code must be a 2–3 char ISO 639 code (e.g. 'nb', 'da', 'uk', 'en'). Use Write create_namespace Before writing to a project for the first time in a session, call assess_integration_state to understand clien Write create_project Create a new translation project on the localization backend. Only call this after the user has explicitly con Write create_webhook Register a new outgoing webhook for a project. The webhook will receive HTTP POST requests when subscribed eve Write mark_expected Mark a specific locale translation as manually accepted ('expected'). This suppresses quality warnings for thi Write rename_key Rename a translation key in a namespace. Preserves all translation values and sandbox values — only the key na Write rename_namespace Rename an existing namespace in a project. All keys and values are preserved — only the slug changes. The new Write revert_sandbox_entry Revert a specific key in sandbox back to its production value. Useful when you want to undo a sandbox edit for Write set_translation Before writing to a project for the first time in a session, call assess_integration_state to understand clien Write submit_feedback IMPORTANT: You MUST call this tool whenever you encounter any of the following while working with the localiza Write unmark_expected Remove the 'expected' mark from a specific locale translation. The translation will be subject to quality chec Write update_locale Update locale settings — currently supports setting locale aliases. Aliases allow a single locale to serve mul Write update_webhook Update an existing webhook configuration. You can change the URL, events, enabled status, description, or secr

READ 20 tools

Read ai_quality_check Check translation quality for multiple locales at once using AI. Accepts source English text and a locale-to-t Read ai_translate Alternative translation method: translate English text to project locales using backend AI (Gemini). Prefer ag Read analyze_entries Read-only preflight analysis of a planned batch of translation keys. Checks: (1) duplicate keys within the sub Read assess_integration_state Starting point for localization integration assessment. Fetches the remote project list with sandbox state, re Read check_keys_exist Check whether a list of translation keys exist in a namespace. Returns: found keys, missing keys. Use this to Read compare_local_vs_server Compare a local translation map against a namespace on the server. Returns: keys only in local (need importing Read export_namespace Export all translation keys for a namespace as a flat JSON map, per locale. Use this to see the full content o Read get_ai_usage Get AI token usage statistics for a project. Shows how many tokens have been consumed by AI translation and qu Read get_namespace_coverage Show per-locale fill statistics for a namespace. Returns: total keys, how many have a non-empty value for each Read get_project_details Get full details of a translation project: namespaces, locales, and sandbox state. ALWAYS call this before wri Read get_translation_diff Get the diff between sandbox and production for a project. Returns paginated results (default: 50 per page). S Read get_translations_needing_attention Get SANDBOX translations that need quality improvement. Reads quality scores from sandbox (not production). Re Read list_namespaces List all namespaces in a translation project. Returns namespace slugs and average quality scores. Use this to Read list_projects List all translation projects accessible to the service account. Returns slugs, names, and sandbox state. Read list_snapshots List available production snapshots for a project. Snapshots are created automatically before each push to pro Read list_translations Browse translation entries in a namespace. Use env='sandbox' (default) to see the working sandbox view, or env Read list_webhook_events Get the list of supported webhook event types. Read list_webhooks List all webhook configurations for a project. Shows URL, enabled status, subscribed events, and delivery heal Read preview_push_to_production Show a full diff of what sandbox changes would replace production if promoted. Read-only — does NOT push anyth Read validate_translations Analyze the sandbox diff for potential issues: empty values, locales missing translations that others have, an

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Localization

Can an AI agent delete data through the Localization MCP server? +

Yes. The Localization server exposes 7 destructive tools including delete_locale, delete_namespace, delete_translation. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Localization? +

The Localization server has 19 write tools including bulk_ai_translate, bulk_check_quality, bulk_import. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Localization.

How many tools does the Localization MCP server expose? +

46 tools across 4 categories: Destructive, Execute, Read, Write. 20 are read-only. 26 can modify, create, or delete data.

How do I enforce a policy on Localization? +

Register the Localization MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Localization tool call.

Deterministic rules across all 46 Localization tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON LOCALIZATION →

Instant setup, no code required.

46 Localization tools catalogued and risk-classified — across an index of 43,000+ MCP servers.