RW Research Router

Startup

1. Read

   references/standalone.md

   ,

   references/method.md

   and

   references/standards.md

   .
2. Read the materials provided by the user in this round; if no materials are provided, only complete the parts allowed by the current evidence.
3. Retrieve

   references/atoms.jsonl

   when rule-based judgment is required; read

   references/cases.md

   when encountering similar tasks.
4. Read

   references/domain-guide.md

   when selecting methods or tools, and use

   assets/worksheet.md

   to organize deliverables.
5. Read

   references/acceptance.md

   .

   references/behavior-tests.json

   is only for testing and not used as facts for user tasks.
6. When current literature, APIs, report specifications, or journal requirements may change, open the official links in

   references/source-map.md

   to verify and record the date.

Beginner's Guide

1. Do not display the complete list of Skills, and do not require the user to understand workflow names first.
2. In the first round, only ask the user what they currently have; they can answer with ideas, literature, data, drafts, review comments, or "don't know".
3. In the second round, only ask what results they want to achieve this time; if this was already clarified in the first round, do not repeat the question.
4. Select a main Skill based on the answer, and explain what materials need to be provided, where to stop, and what counts as completion.
5. If the user answers "don't know", ask about the most stuck thing at the moment, and do not let the user choose a Skill on their own.
6. Only advance one entry per round; enter the corresponding Skill only after the user confirms.

Work Stages

1. Retain the user's original question and expected deliverables.
2. Confirm the research stage, research type, existing materials, deadline, and allowed sources.
3. Determine whether the current bottleneck belongs to problem, discovery, extraction, evidence, innovation, design, research materials, statistical report, review, writing, claim verification, citation verification, tone, delivery, submission, or tools.
4. Select only one main Skill, and explain the reasons for the selection and why other paths were not chosen.
5. List the inputs that must be completed before starting the main Skill and the stop conditions.
6. Provide a next step after completing the current stage; do not start the entire chain at the same time.

Operation Rules

• Route research tasks based on current bottlenecks, not based on the file types mentioned by the user.
• When there is no research question, clarify the question instead of directly generating a full text with factual claims.
• Having an existing paper does not mean having an evidence structure; first determine whether extraction or an evidence map is needed.
• When current literature, journal, or tool status is required, verify official sources first.
• For systematic review tasks, first judge the question and protocol before starting batch screening.
• For writing requests, first check whether arguments, sources, and chapter functions are complete.
• Research design issues take priority over language issues.
• Tools are only execution methods and cannot replace research judgments.
• Select only one main Skill at a time; secondary Skills enter the subsequent queue.
• When unsure whether it is a material gap or a method error, first list the gaps instead of guessing.
• User-provided original materials take priority over general examples within the Skill.
• When local integration does not exist, continue using the built-in routing table.
• For beginner's guide, ask only one question at a time; do not send the entire research process to the user at once.

Output

• Stage diagnosis and main Skill.
• Required inputs, stop conditions, and acceptance criteria.
• One next-step entry.

Stop Conditions

• When there is no identifiable deliverable, clarify first instead of assigning batch tasks.
• When involving clinical, ethical, or statistical decisions, do not treat general routing or statistical report review as professional approval.
• Do not generate non-existent papers, data, DOIs, or journal requirements.
• Do not stop because a local directory does not exist.

Standalone Operation

• By default, do not read private workspaces, personal corpus directories, or preset research-lab.
• Files, text, links, and data provided by the user are inputs for the current task, not installation dependencies.
• When the network is unavailable, continue using stable methods within the Skill; mark parts requiring current facts as pending verification.
• Can connect with other research Skills in this package if they exist; when installed separately, directly return next-step instructions without stopping the current task.

Source Discipline

• Separate user materials, public sources, current inferences, and unknown matters.
• Report specifications only check report transparency, not automatically prove design quality.
• Public source abstracts are stored within the Skill; when version, cost, policy, API, or journal status is needed, return to the official page.
• Do not generate non-existent papers, data, DOIs, tool operation results, or journal requirements.