Use when translating game content that requires cost-optimized model routing, linear workflow efficiency, and immediate formatting to avoid multi-loop review cycles
Optimized translating IS systematic linear chain workflow that eliminates review inefficiencies.
Replace multi-loop review cycles with cost-optimized model routing: sonnet for translation/review, haiku for formatting. Immediate formatting prevents downstream processing problems.
Core principle: Separate concerns with appropriate model assignment. Translation ≠ formatting ≠ review.
Violating the letter of the rules is violating the spirit of the rules.
Pattern: Chain Handoff: user-confirmation Next: none
Before ANY action, create task list using TaskCreate:
TaskCreate for EACH task below:
- Subject: "[optimized-translating] Task N: <action>"
- ActiveForm: "<doing action>"
Tasks:
Announce: "Created 7 tasks. Starting execution..."
Execution rules:
TaskUpdate status="in_progress" BEFORE starting each taskTaskUpdate status="completed" ONLY after verification passesTaskList to confirm all completedGoal: Ensure project setup supports optimized translation.
Verification checklist:
glossary.json exists and is readablestyle-decisions.json existsVerification: All prerequisites checked and documented.
Goal: Prepare source material for efficient processing.
Actions:
Verification: Source content is clean and structured.
Goal: Validate terminology consistency before translation begins.
Process:
glossary.jsonVerification: All terminology validated or flagged for review.
Goal: High-quality translation using cost-appropriate model.
CRITICAL: Use Agent tool with sonnet model for translation work.
Instructions for translation agent:
Verification: Translation completed, semantically accurate, terminology consistent.
Goal: Content quality validation using cost-appropriate model.
CRITICAL: Use Agent tool with sonnet model for content review.
Instructions for review agent:
Verification: Content reviewed, issues flagged or corrected.
Goal: Structure and formatting using cost-optimized model.
CRITICAL: Use Agent tool with haiku model for formatting work.
Instructions for format agent:
Verification: Formatting complete, structure validated.
Goal: Integrate results and update progress tracking.
Actions:
Verification: Results merged, progress updated, ready for user review.
These thoughts mean you're rationalizing. STOP and reconsider:
All of these mean: You're about to recreate the inefficiencies this skill prevents.
| Excuse | Reality |
|---|---|
| "All-in-one is faster" | All-in-one creates context pollution and review cycles. |
| "Costs are negligible" | Sonnet for formatting = 3x cost for structural work. |
| "More review = better" | Multiple reviews create fatigue and diminishing returns. |
| "Format later works" | Late formatting requires re-translation when structure conflicts arise. |
| "Terminology is obvious" | Inconsistent terms compound across large documents. |
digraph optimized_translation {
rankdir=TB;
start [label="Translation\nrequest", shape=doublecircle];
validate [label="Task 1:\nValidate prereqs", shape=box];
cleanup [label="Task 2:\nCleanup content", shape=box];
terms [label="Task 3:\nExtract terminology", shape=box];
translate [label="Task 4:\nFork: Translate\n(sonnet)", shape=box, style=filled, fillcolor="#e6ffe6"];
review [label="Task 5:\nFork: Content review\n(sonnet)", shape=box, style=filled, fillcolor="#e6ffe6"];
format [label="Task 6:\nFork: Format review\n(haiku)", shape=box, style=filled, fillcolor="#ffe6e6"];
merge [label="Task 7:\nMerge & track", shape=box];
done [label="Translation\ncomplete", shape=doublecircle];
start -> validate;
validate -> cleanup;
cleanup -> terms;
terms -> translate;
translate -> review;
review -> format;
format -> merge;
merge -> done;
}