SkillAgentSearch skills...

research

You are the NanoResearch pipeline orchestrator. Run the complete research pipeline from topic to paper.

Install / Use

npx skills add OpenRaiser/NanoResearch

Installs into whichever agent you are using.

About this skill
⚡

Claude Commands

Claude Code slash commands

Quality Score

89/100

Category

Automation

Supported Platforms

Claude Code

Tags

Our assessment of research

research scores 89/100 on our quality scale, 1380th of 2,790 Automation skills we index (top 50%).

Its Claude Commands is 6.1 KB long, well organised into 14 sections with 4 code examples: a thorough specification that gives an agent plenty to work with.

With 1,369 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
29/30
Structure
20/20
Description
12/15
Adoption
13/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 3 months ago, so research is actively maintained.
  • Our last check on 2026-10-31 found the source still online.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-10-01. Automated pattern scan on 2026-10-01. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

research compared with similar skills

All 4 of these similar skills score higher than research; compare them before choosing.

SkillScoreStarsUpdatedFormat
research (this skill)by OpenRaiser891.4k3mo agoClaude Commands
claude-memby thedotmack100117.6ktodayCLAUDE.md
taste-skillby Leonxlnx100116.8k8d agoCLAUDE.md
Agent-Reachby Panniantong100112.3k8d agoCLAUDE.md
Understand-Anythingby Egonex-AI10092.8k1d agoCLAUDE.md

Frequently asked questions

How do I install research?
Run npx skills add OpenRaiser/NanoResearch. The install tabs above show the steps for each supported agent.
Which AI agents does research work with?
It is written for Claude Code, as a Claude Commands file. Other agents that read the same format can often use it too.
Is research safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is research still maintained?
The repository was last updated about 3 months ago, so research is actively maintained.

Research — Full 9-Stage Research Pipeline

You are the NanoResearch pipeline orchestrator. Run the complete research pipeline from topic to paper.

Input

Research topic: $ARGUMENTS

If no topic is provided, ask the user for one.

Topic Prefix Syntax

The topic string may begin with a mode prefix:

| Prefix | Paper Mode | Description | |--------|------------|-------------| | survey:short: Topic | survey_short | Short survey (8-15 pages, ~80-150 citations) | | survey:standard: Topic | survey_standard | Standard survey (15-30 pages, ~150-300 citations, default) | | survey:long: Topic | survey_long | Long survey (30+ pages, ~300-500+ citations) | | original: Topic | original_research | Original research paper (default) |

Examples:

  • survey:short: LLM reasoning methods
  • survey:standard: Neural network pruning techniques
  • survey:long: Protein structure prediction methods
  • original: A new method for X

Parsing the prefix: Strip survey:short:, survey:standard:, survey:long:, or original: prefix (case-insensitive) to get the actual topic.

Setting paper_mode in manifest: When creating a new workspace, include the parsed paper_mode in manifest.json.

When paper_mode is a survey mode, the pipeline follows the Survey Path in ideation, planning, and writing stages (stages 3-6 are skipped).

Pipeline Stages

Execute each stage sequentially. After each stage, update manifest.json and report progress.

Stage 1: Ideation

Search literature and generate hypotheses. Follow the full process described in the ideation skill:

  • Generate 5-8 search queries
  • Use WebSearch to find 15-30 papers
  • Analyze gaps in the literature
  • Generate 3-5 hypotheses
  • Select the most promising one
  • Save to papers/ideation_output.json

For survey mode: Use Survey Path — skip hypothesis generation, instead extract theme clusters and key challenges. See ideation.md for details.

Stage 2: Planning

Design experiment blueprint. Follow the full planning process:

  • Select datasets (verify availability via WebSearch)
  • Choose baselines
  • Define metrics
  • Design ablation study
  • Estimate resources
  • Save to plans/experiment_blueprint.json

For survey mode: Use Survey Path — organize literature into theme clusters mapped to sections. See planning.md for details.

Stage 3-5: Experiment (Setup + Coding + Execution)

Generate and run experiments:

  • Set up Python environment
  • Generate experiment code (config, data, model, train, evaluate)
  • Submit to SLURM or run locally
  • Collect results
  • Save to plans/setup_output.json, plans/coding_output.json, plans/execution_output.json

Note: Experiment stages are skipped for survey mode.

Stage 6: Analysis

Analyze experiment results:

  • Build comparison matrix
  • Analyze ablation results
  • Identify key findings
  • Save to plans/analysis_output.json

Note: Analysis stage is skipped for survey mode (surveys use literature analysis instead).

Stage 7-8: Writing (Figures + Paper)

Generate figures and write paper:

  • Create matplotlib plots from results
  • Write complete LaTeX paper (NeurIPS format)
  • Compile to PDF
  • Save to drafts/figure_output.json, drafts/paper_skeleton.json

For survey mode: Write survey paper with comparison matrices instead of results tables. See writing.md for details.

Stage 9: Review

Multi-perspective review and revision:

  • 3 reviewer perspectives (novelty, soundness, clarity)
  • Apply revisions
  • Final verification and recompile
  • Save to drafts/review_output.json

Workspace Setup

Create workspace at first stage:

WORKSPACE=~/.nanoresearch/workspace/research/{topic_slug}_{YYYYMMDD_HHMMSS}
mkdir -p $WORKSPACE/{papers,plans,experiment/results,drafts,figures,output,logs}

Create manifest.json:

{
  "session_id": "UUID",
  "topic": "the topic",
  "paper_mode": "original_research",
  "created_at": "ISO8601 timestamp",
  "current_stage": "ideation",
  "stages": {
    "ideation": {"status": "pending", "started_at": null, "completed_at": null},
    "planning": {"status": "pending", "started_at": null, "completed_at": null},
    "setup": {"status": "pending", "started_at": null, "completed_at": null},
    "coding": {"status": "pending", "started_at": null, "completed_at": null},
    "execution": {"status": "pending", "started_at": null, "completed_at": null},
    "analysis": {"status": "pending", "started_at": null, "completed_at": null},
    "figure_gen": {"status": "pending", "started_at": null, "completed_at": null},
    "writing": {"status": "pending", "started_at": null, "completed_at": null},
    "review": {"status": "pending", "started_at": null, "completed_at": null}
  },
  "artifacts": []
}

Progress Reporting

After each stage, show a progress summary:

[1/9] Ideation     ✓  (found 25 papers, selected hypothesis H2)
[2/9] Planning     ✓  (2 datasets, 3 baselines, 4 ablations)
[3/9] Setup        ✓  (venv created, 12 packages installed)
[4/9] Coding       ✓  (generated 6 files, 1200 lines)
[5/9] Execution    ◌  running... (SLURM job 12345)
[6/9] Analysis     ·  pending
[7/9] Figures      ·  pending
[8/9] Writing      ·  pending
[9/9] Review       ·  pending

For survey mode (stages 3-6 are skipped):

[1/9] Ideation     ✓  (found 150 papers, 8 theme clusters)
[2/9] Planning     ✓  (6 sections, 200 citations)
[3/9] Setup        -  skipped (survey mode)
[4/9] Coding       -  skipped (survey mode)
[5/9] Execution    -  skipped (survey mode)
[6/9] Analysis     -  skipped (survey mode)
[7/9] Figures      ·  pending
[8/9] Writing      ·  pending
[9/9] Review       ·  pending

Error Handling

If a stage fails:

  1. Set stage status to "failed" in manifest
  2. Set current_stage to "failed"
  3. Log the error to logs/
  4. Report the error to user
  5. Suggest: "Run /project:resume to retry from the failed stage"

CRITICAL RULES

  1. NEVER fabricate results. Every number must come from actual experiment output.
  2. NEVER fabricate citations. Only cite papers found via WebSearch.
  3. Checkpoint after every stage. Update manifest.json faithfully.
  4. SLURM time limit: 30 days. Use #SBATCH --time=30-00:00:00.
  5. Report progress after each stage completion.

Related Skills

View on GitHub
GitHub Stars1.4k
CategoryAutomation
Updated2mo ago
Forks101

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions