agentleFS
wangyi0403wangyi0403/dl-research-workflow/.agents/skills/paper-plan/SKILL.md

paper-plan

Skill4 starsChanged 41 days agoMITView on GitHub

Plan an evidence-led paper from verified claims, results, and venue constraints. Use before drafting to map its narrative, sections, figures, limitations, and evidence.

12 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use daily-arxiv, exp-eval and 9 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LambenthanSkill

    daily-arxiv

    每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新

    Also in 5 projects that use this

  • LambenthanSkill

    exp-eval

    实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges

    Also in 5 projects that use this

  • raja21068Skill

    deepxiv

    Search and progressively read open-access academic papers through DeepXiv. Use when the user wants layered paper access, section-level reading, trending papers, or DeepXiv-backed literature retrieval.

    Also in 4 projects that use this

  • raja21068Skill

    experiment-bridge

    Workflow 1.5: Bridge between idea discovery and auto review. Reads EXPERIMENT_PLAN.md, implements experiment code, deploys to GPU, collects initial results. Use when user says \"实现实验\", \"implement experiments\", \"bridge\", \"从计划到跑实验\", \"deploy the plan\", or has an experiment plan ready to execute.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-eval

    Pilot result evaluation — read pilot results, apply success criteria, update idea page (pilot_result, failure_reason if failed), generate PILOT_VERDICT_REPORT. Called by /ideate Phase 5 after /exp-pilot-run.

    Also in 4 projects that use this

  • skyllwtSkill

    exp-pilot-run

    Pilot experiment execution — read Pilot Spec YAML, write pilot code, run experiment(Confirm with the user before operation and require the applicant to conduct manual inspection), return results. Called by /ideate Phase 5. Does NOT modify wiki pages or judge pass/fail.

    Also in 4 projects that use this

  • +5 more from Lambenthan

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 11 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use exact citation metadata only after verification.

    line 30
  2. Distinctive:

    Do not write the full manuscript, alter results, or submit material as part of planning.

    line 31
  3. Very specific:

    When an engineering method paper needs a concrete introduction pattern, consult references/intro-flowchart.md as an optional example.

    line 35

“Common” is measured: kept by at least 6 in 10 of the 11 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Paper Plan

Build the narrative before drafting. The plan must make it possible to trace every central statement to evidence and to distinguish the contribution from its implementation.

Inputs and output

Read AGENTS.md, docs/03_idea_report.md, docs/04_methodology.md, docs/05_experiment_plan.md, docs/06_result_tables.md, docs/08_analysis.md, verified citations, and the target venue's actual author guidance. Read docs/10_figure_report.md only if it already contains an earlier figure inventory; Stage 3A must not depend on a later Stage 3B artifact. Update docs/09_paper_plan.md and its Gate D record.

Workflow

  1. Select the paper type, target reader and central claim. State the comparison or knowledge gap, tested conditions, evidence status and maximum permissible wording. Method, measurement, benchmark, theory, negative-result and systems contributions may require different evidence and ordering.
  2. Build the contribution chain: problem → evidence gap → mechanism or design principle → method as a testable implementation → decisive evidence → scope and limitations.
  3. For engineering research, make the chain explicit: real field problem and consequence → why existing engineering paths are inadequate → reusable measurement/physical/inference question → supported new knowledge or principle → concrete method → separate engineering and scientific validation. Do not frame a VLM, model family, or a collection of modules as the contribution by itself.
  4. Organize necessary supporting claims around the central answer. For each, identify the reader question, required evidence, strongest concrete alternative, exact readout and any open evidence gap; reuse the project's existing Claim and analysis records. Order sections by the reader's dependencies rather than the project chronology. If the project provides docs/WRITING_STANDARD.md, instantiate its user-specific budgets in docs/09_paper_plan.md; current user instructions, approved structure and verified venue constraints take precedence. Do not invent content to fill quotas.
  5. Select representations by reader task: prose for a simple fact, a table for exact lookup, a plot for patterns, an equation for a relation, and an algorithm for consequential sequence/state. Pair them only for distinct jobs. Put decisive evidence and interpretation-critical conditions in the main text; the supplement can hold exhaustive reproducibility detail. Quantitative displays must be reproducible, and method names agree with the methodology.
  6. Prewrite the limitations and permitted claims. Preserve evidence-supported boundaries, including coverage, weak signals, data quality, causal scope, and generalization limits; do not add generic disclaimers or promotional future work.
  7. Conduct an independent local review using $research-quality-gate or a clearly separated critic pass. Check logical flow, claim-evidence alignment, missing decisive tests, closest-work positioning, figure plan, and venue feasibility. Record minimum fixes in the Gate D section.

Rules

  • A good outline is not an abstract expanded into headings. Each central claim needs a known evidence location.
  • A task-performance gain does not automatically establish a mechanism explanation or transfer claim.
  • A missing experiment is proposed only when positive, negative or inconclusive outcomes can change a claim or decision. Narrow a claim when that is the sufficient evidence-based repair; do not add every possible control or fix a paper around a favorable post hoc subset.
  • Title, abstract, introduction, principal displays and conclusion express the same supported promise at different levels of detail. A scoped trade-off or a negative finding can be central when the actual evidence supports it.
  • Use exact citation metadata only after verification; leave unresolved sources marked for verification rather than generating BibTeX from memory.
  • Do not write the full manuscript, alter results, or submit material as part of planning.

Report the narrative spine, Claim/Chain-to-section matrix, figure/table plan, limitations, unresolved evidence, and Gate D decision.

When an engineering method paper needs a concrete introduction pattern, consult references/intro-flowchart.md as an optional example. Use it only where it fits the selected paper type and evidence; it does not set contribution counts or a mandatory section structure.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.