Use this skill when the user wants the agent to judge whether an answer is really backed by the document, not just whether it sounds correct. Trigger it for requests like "check if the citation really supports the claim," "make examples of sneaky hallucinations," "verify grounded answers sentence by sentence," or "test whether the model invents support from the source." It is ideal for legal research outputs, RAG answers, summaries, and any document-grounded generation setting where false support is as dangerous as an outright fabrication.
[Case 1]
[Case 2]
[Case 3]
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:
Phase 1: Environment Exploration
Read the exploration guidelines to discover raw knowledge seeds:
references/EXPLORATION.md
Phase 2: Trajectory Selection
Once Phase 1 is complete, read the selection criteria to evaluate the trajectory:
references/SELECTION.md
Phase 3: Data Synthesis
Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data:
references/SYNTHESIS.md