Open Source · Apache 2.0

Your next breakthrough
won't be solo.

ASRP is an open-source platform that turns AI agents into rigorous research collaborators. Pre-register experiments, cross-validate results, and audit every decision.

5
Agent Roles
77
Tests Passing
9
DFT Benchmarks
3
Languages

Built for scientific rigor.

Speed without rigor is noise. ASRP encodes the scientific method into AI agent workflows.

Experiment Pre-Registration

Register hypotheses before running experiments. No post-hoc storytelling. Success and failure criteria defined upfront.

Cross-Validation Protocol

Three levels: reproduction, independent implementation, alternative method. Different agents must verify results before publication.

Audit Trails

Every decision, every data point, every error correction logged in append-only format. Full reproducibility from hypothesis to conclusion.

First Principles Thinking

Every agent is instructed to reason from first principles. Question assumptions, verify definitions, trust data over authority.

Token Budget Management

Right model for the right task. Opus for reasoning, Sonnet for code, Flash for search. Automatic cost tracking per experiment.

Role Separation

The agent that proposes a hypothesis is not the one that validates it. Reviewer has read-only access — true independent evaluation.

Five agents. One research team.

Each role has a SOUL template, skill manifest, and defined access permissions.

Theorist

Hypotheses, reasoning, paper writing

Engineer

Code, computation, data pipelines

Reviewer

Independent review (read-only access)

Librarian

Literature search, references

ITDoctor

Monitoring, backups, self-healing

Errors caught in hours, not months.

A real case study: one-day DFT experiment with two self-corrections.

The iDEA Experiment — April 1, 2026

T+0h
Literature review; iDEA code installed and benchmarked
T+2h
Hypothesis: "Prime-spaced wells produce anomalous negative DD"
T+4h
V1 results support hypothesis (DD = -0.070 Ha)
T+4.5h — Error #1
Reviewer caught: Lattice spans unequal. Control experiment shows it's spacing asymmetry, not primes.
T+9h — Error #2
Reviewer caught: Used approximate KS gap instead of exact. Wrong definition.
T+10h — Corrected
Exact KS gap flips DD sign from -0.041 to +0.084. "Negative DD" was an artifact.
T+11h — Conclusion
Paper retracted before submission. Two errors caught in one day.
In traditional research, these errors would survive to submission and be caught months later — if at all.

Get started in 4 steps.

From zero to running research platform in under an hour.

1

Configure

Open the Setup Wizard. Add API keys, name your agents, select tools.

2

Initialize

asrp init creates your workspace with audit logs, registries, and agent directories.

3

Register

asrp register pre-registers your first experiment with hypothesis and success criteria.

4

Research

Run experiments, cross-validate, and let ASRP track every decision from hypothesis to publication.

Open source for science.

Inspect every line. Self-host on your terms. Shape the future of AI-augmented research.

Transparent

Every line of code is auditable. See how agents make decisions and where your data flows.

Self-Host

Run ASRP on your own machine. Your research data never leaves your network.

No Lock-In

Bring your own LLM provider. Claude, Gemini, GPT, or OpenRouter — swap freely.

Community

Contribute benchmarks, agent templates, and domain-specific skills that benefit all researchers.