What this resource does
Core uses
About this resource
NanoResearch combines research-planning Skills with a Python CLI that can search literature, plan experiments, run local or SLURM jobs, analyse results, create figures, and export a LaTeX paper workspace.
This page groups representative academic components by task; review the repository for the complete inventory.
Research ideation and planning
Searches literature, identifies a research direction, and turns it into an experiment plan with datasets, baselines, metrics, and ablations.
- Typical inputs
- Research topic; Target venue, constraints, and available compute
- Output
- A literature-grounded idea record and experiment blueprint.
Best for: Computational researchers who need a traceable plan before generating code.
Components for this task
nanoresearch-ideation
Handles literature search, research-gap exploration, and hypothesis development.
Suggested workflow requests (2)
Run the ideation stage for adaptive sparse attention mechanisms, then propose a testable experiment plan.
Create an experiment plan for my approved research topic, including datasets, baselines, metrics, and ablations.
Adapted into complete requests from official trigger wording in README.md.
Experiment execution and analysis
Generates experiment code, runs it locally or through SLURM, monitors logs, and records real metrics and artifacts.
- Typical inputs
- Approved experiment plan; Datasets, environment, compute limits, and stopping rules
- Output
- Runnable code, logs, metrics, manifests, and analysis artifacts.
Best for: Running bounded machine-learning experiments with recoverable workspace state.
Component for this task
Figure and paper production
Uses recorded experiment artifacts to create figures and assemble a LaTeX manuscript package.
- Typical inputs
- Verified experiment outputs; Paper format and writing constraints
- Output
- Figures, LaTeX source, references, and an exportable paper workspace.
Best for: Preparing a first manuscript draft after the underlying results have been checked.
Component for this task
nanoresearch-writing
Creates paper figures and writing artifacts from recorded experiment evidence.
Use boundaries
Limits and checks
Local and cluster execution
It may consume significant compute or change the research workspace.
Use a dedicated environment, explicit budgets, version control, and dry-run checks.
Result interpretation
A software or design error can propagate into figures and manuscript claims.
Review code, rerun key results, and compare every claim with raw artifacts.
External model and image APIs
Costs, retention, and data-handling terms depend on those providers.
Check provider terms and avoid sending confidential material without approval.
More boundaries
- Do not use the full pipeline when you cannot provide a bounded computational question, usable data, and enough compute.
- Generated code and automatic retries do not establish that an experiment is methodologically valid or reproducible.
- A generated paper package is a draft and still needs source, result, authorship, and venue-policy checks.
Technical details
- Resource type
- Toolkit
- Author or maintainer
- OpenRaiser
- Latest release
- Static Assets (assets)
- Source last updated
- 26 May 2026
- Last verified
- 15 Jul 2026
- Documented applications
- Claude Code, Codex, NanoResearch CLI
- Documented AI models
- DeepSeek-V3.2, Claude Sonnet 4.6, GPT-5.5, gpt-image-2
- Licence
- MIT
- Access
- Publicly available
- Additional costs
- Platform terms or usage limits may apply. API usage fees may apply for selected components. External services, software, compute, or data access may have separate costs.
- Skill instruction language
- English
- Documentation language
- English
- Repository languages
- Python, Markdown
- Dependencies
- Python and pip; Git; OpenAI-compatible model API for the CLI route; Local GPU or SLURM access for compute stages; Component-specific datasets and packages
- Review status
- Not tested
Continue exploring