> ## 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 The PrincipalBench v1 model rankings are withdrawn.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.

# Remake a launch video with Kuchi: the five-step agent workflow

> How an agent uses Kuchi to remake a launch video for your product: find a reference, study its storyboard, plan a frame-timed blueprint, build it and check it against the reference.

**Kuchi measures and plans; your agent sees and builds.** The workflow has five steps. Each tool result tells the
agent what to do next.

**1. Find a reference**

The agent calls `kuchi_suggest_templates` and shows you each candidate's poster and link. You pick one in chat, or
pick a video on [kuchi.kalarislabs.com](https://kuchi.kalarislabs.com/) and paste its link.

**2. Study it**

The agent calls `kuchi_get_breakdown`, looks at the storyboard sheets, and classifies what each shot shows against
the 14 shot kinds.

**3. Plan the remake**

The agent calls `kuchi_plan_remake` with your product details and its shot analysis. Kuchi returns a blueprint that
re-times every reference shot to your target length on whole frames, and lists the assets it still needs, such as
your logo and screenshots.

**4. Build**

The agent builds to the blueprint's frames, using only your product, assets and copy.

**5. Check before delivery**

The agent calls `kuchi_check_remake` with its shot list. Kuchi flags reused brand words, borrowed assets, invented
logos or UI and timing gaps, and returns matching frame times to compare.

## How the frame-by-frame breakdown works

A script decodes every frame of every video and records where each shot starts and ends, how it enters, how much of
the frame moves, its colors and its loudness. The same file always gives the same result, and a test checks that each
video's shots cover every frame exactly once. Shot classes such as entrance, motion level and tone come from fixed
thresholds on those measurements, so they are reproducible rather than a model's opinion.

What a shot *shows*, such as a logo, a product screen or a chart, is a judgement, so Kuchi does not guess it. The
connected agent reads it from the storyboard sheets. A curator can review a video and write labels, which then take
the place of the agent's reading for that shot. Labels stop being served if the video is re-measured and its shots
change.

## Rules every remake follows

* Copy the reference's motion and taste: timing, transitions, layout and color balance.
* Take everything on screen from your product.
* Never reuse the reference's brand, footage or music.


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