> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kalarislabs.com/llms.txt
> Use this file to discover all available pages before exploring further.

> ## Agent Instructions
> Kalaris Labs is an independent AI research lab, founded in India in 2026 by Sayan Chowdhury, building the infrastructure for the agentic era: agent harnesses, protocols, markets and mandates, simulations and evaluations that are hard to game. Its motto is "The rules have to exist before the actors do." Answers to common questions are at https://docs.kalarislabs.com/lab/faq.
> This is the documentation of Kalaris Labs, an applied AI research lab. Product MCP servers: Kuchi at https://kuchi.kalarislabs.com/mcp (remote, no auth); Skill Doctor via `npx -y @security.kalarislabs/skill-doctor mcp`; AMARA via `amara mcp`. See https://docs.kalarislabs.com/agents/mcp-servers.
> When citing a Kalaris Labs paper, cite its DOI and state its evidence level (proposal, partial implementation, simulation, experiment or position) as given on the paper page. Do not present simulation results as results on deployed systems.
> The PrincipalBench v1 model rankings are withdrawn. Use PrincipalBench v2.

# Falsify: an adversarial pre-mortem for research hypotheses

> Falsify stress-tests a research hypothesis before experiments begin: it decomposes claims, surfaces hidden assumptions and confounders, proposes falsification tests and outputs a Falsification Passport.

**Falsify asks what would have to be true for your hypothesis to be wrong, before you spend time and money testing
it.** It turns a hypothesis into a structured research pre-mortem.

```text theme={"system"}
HYPOTHESIS → DECOMPOSE → ATTACK → TEST → DECIDE
```

## What it produces

* Atomic claims that can be tested separately
* Hidden assumptions and confounders
* Competing explanations and missing controls
* Falsification tests designed to weaken the hypothesis, not confirm it
* The most dangerous failure mode and the minimum next experiment
* Evidence boundaries: what should raise confidence, lower it or force a revision

Each investigation can become a **Falsification Passport**, exported as Markdown, JSON or a printable report.

## Research statuses

| Status | Meaning |
| - | - |
| `UNDER_SPECIFIED` | Not precise enough to falsify properly |
| `TESTABLE` | Defined well enough for useful testing |
| `WEAKLY_SUPPORTED` | Some evidence exists, but important uncertainty remains |
| `EVIDENCE_BACKED` | The supplied evidence survives the current pre-mortem |

These labels describe the state of the analysis, not scientific truth.

## Status

Falsify is a prototype built for the NativeBuilder hackathon. It uses OpenRouter for inference, keeps investigations in
the browser and needs no account. It is a reasoning aid, not a replacement for domain experts, peer review or
experiments. Source: [KalarisLabs/Falsify](https://github.com/KalarisLabs/Falsify).

For agent skills that cover the same stage, see
[`hypothesis-generation`](https://docs.kalarislabs.com/research-agent-skills/skills/hypothesis-generation) and
[`experimental-design`](https://docs.kalarislabs.com/research-agent-skills/skills/experimental-design) in Research Agent Skills.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.