English version: See the corresponding English skill.
When to use
- When you need to write more complete and prioritized test cases based on multi-source materials.
- When you need stronger traceability relationships and risk awareness than the basic edition.
Script capability boundaries
Optional scripts/run_*.py Only prepare parsed input materials or source candidates, output NOT_ASSESSED, Do not call the model, do not execute the main Prompt, and do not generate semantic conclusions, risk priorities, complete test cases, or approval. Formal analysis is executed by the Agent according to the main Prompt; when you need to run scripts, first read references/script-contract.md。
Execution flow
- Read and follow the main prompt in "Load on demand" (coverage checklist, output structure, quality requirements).
- Only supplement the project context that truly affects the results: scope, environment, constraints, risks, dependencies, expected outputs.
- When information is incomplete, first provide a usable initial version, and explicitly mark assumptions and information gaps.
- Default Markdown; switch only when the user specifies another format.
Core constraints
- Prioritize by risk/business impact; do not spread them evenly.
- Write "confirmed facts" and "current assumptions" separately.
- Do not fabricate interface, field, environment, or root cause details that the user did not provide.
- Results must be actionable: scenarios specific, prioritized, and able to guide the next step.
Load on demand
- Before producing output, you must read and follow
prompts/testcase-writer-plus.md(minimum coverage checklist, output structure, quality requirements). - When you need to apply a ready-made template: read the matching template in
output-templates/, and do not invent conflicting structures. - When the user wants examples or benchmarking against existing assets: read the relevant examples in
examples/. - When you need in-depth materials such as framework specifications, troubleshooting, or report schemas: only read the files in
references/that are relevant to the current problem; do not read through the entire directory. - When you need format conversion or auxiliary validation: prefer using the existing scripts in
scripts/rather than rewriting them. - When you need to evaluate/regress this skill: use
evals/, and validate and run with skill-up.
Self-check before delivery
- The output structure of the main prompt has been followed
- Minimum coverage concerns: scope, priority grouping, traceability, preconditions, test data, positive scenarios, exception scenarios, boundary scenarios… (details are subject to the main prompt)
- The minimum checklist has been covered, or it has been indicated why it is omitted
- High-risk items have clear priorities
- No details were fabricated that the user did not provide
- Assumptions and information gaps have been marked
Common pitfalls
- When the scope and context are unclear, do not pretend that it is already complete and usable.
- Do not write all items as equally important.
- Do not skip assumptions and information gaps.
- Do not output large sections of vague theory unrelated to the current toolchain.