BluTranslate
Translate files (PDF, DOCX, PPTX) to any language using the Bluente Translation API. Asks for API key, source files, target language, and output location.
Like a lobster shell, security has layers — review code before you run it.
License
SKILL.md
Translate files using the Bluente Translation API.
Step 1: Collect user inputs
Ask the user using AskUserQuestion. IMPORTANT: Every question MUST have at least 2 options (AskUserQuestion requires min 2 options per question). Use the user's "Other" free-text option for custom input.
Ask all 4 questions in a single AskUserQuestion call:
-
API Key — "What is your Bluente API key? (starts with sk-, get one at https://translate.bluente.com/settings/?type=apiKeys)"
- Options: "I have my key ready" (description: "I'll type it in the Other field below"), "I need to get a key" (description: "Visit https://translate.bluente.com/settings/?type=apiKeys")
- The user will type their actual key in the "Other" free-text field.
-
Source path — "What is the path to the file or folder you want to translate?"
- Options: "Current directory" (description: "Use the current working directory"), "Enter path" (description: "I'll specify a custom path in the Other field")
-
Target language — "What language do you want to translate to?"
- Options (pick 4, use the most common): "Arabic" , "French", "Spanish", "German"
- The user can pick "Other" for any other language.
-
Output path — "Where should the translated files be saved?"
- Options: "Same folder as source" (description: "Save next to the originals"), "Enter path" (description: "I'll specify a custom path in the Other field")
If the user selected "I have my key ready" but didn't type a key in "Other", ask them to provide the key in a follow-up message (don't use AskUserQuestion again for this — just ask in plain text).
Step 2: Look up the correct language code
Use this known mapping of common language codes to avoid an extra API call:
- Arabic =
ara, English =en, French =fra, Spanish =spa, German =de, Chinese =zho, Japanese =ja, Korean =ko, Portuguese =pt, Italian =it, Turkish =tr, Russian =ru, Hindi =hi
If the user picks a language NOT in this list, call the supported languages endpoint:
GET https://api.bluente.com/api/20250924/blu_translate/supported_languages
Header: Authorization: Bearer <API_KEY>
Response format: {"message": "success", "code": 0, "data": [{"name": "French", "svcCode": "fra", ...}, ...]}
Search data array for the matching language and use its svcCode.
Step 3: Detect source files
- If the source path is a single file, translate just that file.
- If the source path is a folder, find all translatable files in it (pdf, docx, pptx, xlsx, etc.)
- If no translatable files found, tell the user and stop.
Step 4: Translate each file
Write and execute a single Python script that handles the entire translation workflow for all files. This avoids shell variable issues (e.g. status is read-only in zsh) and is more reliable than chaining curl commands.
The Python script should:
import requests, time, sys, os
API_KEY = "<API_KEY>"
BASE = "https://api.bluente.com/api/20250924/blu_translate"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}
FILES = [<list of absolute file paths>]
FROM_LANG = "en" # or other source language code
TO_LANG = "<target svcCode>"
OUTPUT_DIR = "<output directory>"
os.makedirs(OUTPUT_DIR, exist_ok=True)
results = []
for filepath in FILES:
filename = os.path.basename(filepath)
ext = filename.rsplit(".", 1)[-1].lower()
to_type = {"pdf": "pdf", "docx": "word", "doc": "word", "pptx": "pptx"}.get(ext, ext)
print(f"Translating: {filename}")
# Upload
with open(filepath, "rb") as f:
r = requests.post(f"{BASE}/upload?glossary=0&engine=3", headers=HEADERS, files={"file": f})
resp = r.json()
if resp.get("code") != 0:
print(f" Upload failed: {resp}")
results.append((filename, False, "Upload failed"))
continue
task_id = resp["data"]["id"]
print(f" Task ID: {task_id}")
# Wait for processing
for _ in range(40):
r = requests.get(f"{BASE}/check?entry=get_status&id={task_id}", headers=HEADERS)
st = r.json()["data"]["status"]
if st == "SERVICE_PROCESSED":
break
if st == "ERROR":
print(f" Processing error")
results.append((filename, False, "Processing error"))
break
time.sleep(3)
else:
results.append((filename, False, "Processing timeout"))
continue
if st == "ERROR":
continue
# Start translation
r = requests.post(f"{BASE}/translate?engine=3", headers=HEADERS,
json={"id": task_id, "action": "start", "from": FROM_LANG, "to": TO_LANG})
resp = r.json()
if resp.get("code") != 0:
print(f" Translation start failed: {resp.get('message')}")
results.append((filename, False, f"Start failed: {resp.get('message')}"))
continue
# Wait for translation
for _ in range(120):
r = requests.get(f"{BASE}/check?entry=get_status&id={task_id}", headers=HEADERS)
st = r.json()["data"]["status"]
if st == "READY":
break
if st == "ERROR":
print(f" Translation error")
results.append((filename, False, "Translation error"))
break
time.sleep(5)
else:
results.append((filename, False, "Translation timeout"))
continue
if st == "ERROR":
continue
# Download
r = requests.get(f"{BASE}/download?id={task_id}&to_type={to_type}", headers=HEADERS)
out_path = os.path.join(OUTPUT_DIR, filename)
with open(out_path, "wb") as f:
f.write(r.content)
print(f" Saved: {out_path}")
results.append((filename, True, out_path))
# Summary
ok = [r for r in results if r[1]]
fail = [r for r in results if not r[1]]
print(f"\nDone: {len(ok)} succeeded, {len(fail)} failed")
for name, _, path in ok:
print(f" OK: {name} -> {path}")
for name, _, err in fail:
print(f" FAIL: {name}: {err}")
Fill in the API_KEY, FILES, FROM_LANG, TO_LANG, and OUTPUT_DIR variables from the user's inputs, then run the script with python3. Set a timeout of 600000ms (10 min) on the Bash tool call.
Step 5: Report results
Tell the user:
- How many files were translated successfully
- Where the translated files are saved
- Any errors that occurred
API Reference
- Docs: https://www.bluente.com/docs
- Base URL:
https://api.bluente.com/api/20250924/blu_translate - Auth:
Authorization: Bearer <API_KEY> - Engines:
3= LLM,4= LLM Pro - Source language (
from) is required — it does NOT auto-detect - Language codes are custom — use the mapping above or verify via supported_languages endpoint
$ARGUMENTS
Files
1 totalComments
Loading comments…