From 科研写作助手 (Research Writing Assistant)
Plans experiment protocols, result tables, mock data, evaluation gates, method traceability, and table schemas for research papers before real results exist.
How this skill is triggered — by the user, by Claude, or both
Slash command
/research-writing-assistant:experiment-results-planningThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill designs the experiment/result layer before final metrics exist. It may generate mock planning data, but never presents mock data as real experimental evidence.
This skill designs the experiment/result layer before final metrics exist. It may generate mock planning data, but never presents mock data as real experimental evidence.
Before writing Results or Discussion, create:
plan/experiment-protocol.mdplan/review/method-experiment-traceability.mdtables/table-schema.mdfigures/data-manifest.mdmock_* filesThe protocol must include:
Each contribution in Introduction must map to at least one experiment or limitation note.
Use these gates in plan/stage-gates.md for result-heavy papers:
plan/review/method-experiment-traceability.md.tables/table-schema.md, figures/data-manifest.md, and data files.plan/review/<section>-peer-review.md.Create:
| Contribution | Method module | Experiment | Table/Figure | Allowed claim | Evidence status |
|---|---|---|---|---|---|
Do not let a contribution survive in Introduction if no experiment, limitation note, or future-work boundary supports it.
Mock or synthetic values are allowed only for planning figures and table layout.
Rules:
mock_ or synthetic_.PLANNING DATA - replace before submission.[待真实实验替换].For each table, define:
| Table | Purpose | Rows | Metrics | Data source | Replacement owner |
|---|
Do not create a table unless it supports a claim in the manuscript.
Recommended table fields include mean ± std or confidence intervals when repeated runs are expected. Record aggregation rules in tables/table-schema.md.
Data figures must go through figures-python:
figures/data-manifest.md.figures/<section>/<figure>.py.Model architecture and flow diagrams use figures-diagram prompts instead of synthetic data plotting.
For real data:
The method achieves X under condition Y, compared with baseline Z. The improvement is mainly associated with [module], while [failure case] remains visible in [metric].
For planning data:
[待真实实验替换] This paragraph will compare Table N after real experiment logs are inserted.
Never leave "experiment purpose", "discussion prompt", or "table position" instructions inside final chapter files.
npx claudepluginhub norman-bury/research-writing-skillCreates a detailed, reproducible research and experiment plan from a validated idea. Steps break goals, data, methods, ablation/sensitivity/robustness tests, significance checks, scheduling, risk, and cost estimates into actionable entries.
Orchestrates full research pipeline from Brainstorming to Reporting via Planning, Implementation, Testing & Visualization phases with user checkpoints. Configurable for physics, AI/ML, statistics, math domains, depth, and agent personas.
Writes biology lab reports using IMRaD format with correct statistical reporting, raw data organization, and reproducible methods sections.