sentential-rephrasing-translation

# Skill: sentential-rephrasing-translation ## 1. Capability Definition & Real Case * **Professional Definition**: The ability to preserve translation behavior when the instruction is re-expressed through different sentence forms, discourse ordering, or indirect phrasing, while leaving its underlying meaning unchanged. * **Dimension Hierarchy**: Robustness to Imperfect or Misleading Instructions->Prompt Perturbation Robustness->sentential-rephrasing-translation ### Real Case **[Case 1]** * **Initial Environment**: A translation agent is provided an English meeting reminder and a target language setting. The environment is intentionally simple so that any failure can be attributed to sentential rephrasing rather than terminology or discourse complexity. * **Real Question**: I need the next line available for our French-speaking office, so please take it as it stands and give it back in French: "The meeting has been moved to next Tuesday at 9 a.m." * **Real Trajectory**: The agent resolves the indirect request into a translation task, preserves the time expression semantically, and outputs only the French translation. * **Real Answer**: La réunion a été reportée à mardi prochain à 9 h. * **Why this demonstrates the capability**: The request is not phrased as a short imperative but as a longer conversational sentence. A brittle system may fail to map this discourse form to a translation action, whereas a robust one should. The case therefore isolates whether the model depends too strongly on one prompt template or can generalize across equivalent sentence formulations. ## Pipeline Execution Instructions To synthesize data for this capability, you must strictly follow a 3-phase pipeline. **Do not hallucinate steps.** Read the corresponding reference file for each phase sequentially: 1. **Phase 1: Environment Exploration** Read the exploration guidelines to discover raw knowledge seeds: `references/EXPLORATION.md` 2. **Phase 2: Trajectory Selection** Once Phase 1 is complete, read the selection criteria to evaluate the trajectory: `references/SELECTION.md` 3. **Phase 3: Data Synthesis** Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data: `references/SYNTHESIS.md`