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 The PrincipalBench v1 model rankings are withdrawn.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

# Agent economies: mechanism design for markets of AI agents

> Overview of the Kalaris Labs agent economies series: context pricing, Liquidity Coupling, the Orchestration Trilemma, the Agent Coupling Protocol and economic defences against prompt injection.

**When AI agents hire other agents, three things break that human markets take for granted: context is finite,
checking a counterparty takes longer than deciding, and one default can cascade faster than a person can react.**
This five-paper series designs economic mechanisms for each pressure and combines them into one protocol. All results
in the series come from simulation.

## Read in this order

**1. Price context: The Marginal Utility of Context**

An exponential barrier price on context use makes a rational agent compress memory early.
[Read the paper](https://docs.kalarislabs.com/research/papers/marginal-utility-of-context.md).

**2. Stop default cascades: Liquidity Coupling**

Escrowing a fraction α of each contract makes insolvency cascades die out once α ≥ 1 − 1/λ.
[Read the paper](https://docs.kalarislabs.com/research/papers/liquidity-coupling.md).

**3. Name the problem: The Orchestration Trilemma**

Context, trust and solvency constraints together defeat mechanisms built for human markets.
[Read the paper](https://docs.kalarislabs.com/research/papers/orchestration-trilemma.md).

**4. Combine the mechanisms: The Agent Coupling Protocol**

A ten-stage protocol that weighs reliability more heavily as an orchestrator's context fills.
[Read the paper](https://docs.kalarislabs.com/research/papers/agent-coupling-protocol.md).

**5. Apply them to security: economic safeguards against prompt injection**

The same mechanisms make a successful indirect prompt injection cost more than it is worth.
[Read the paper](https://docs.kalarislabs.com/research/papers/economic-safeguards-indirect-prompt-injection.md).

## How the mechanisms map to the pressures

| Pressure | Mechanism | Paper |
| - | - | - |
| Finite context window | Exponential context barrier φ(c) | [Marginal Utility of Context](https://docs.kalarislabs.com/research/papers/marginal-utility-of-context.md) |
| Defaults spread faster than humans react | Escrowed stake (Liquidity Coupling) | [Liquidity Coupling](https://docs.kalarislabs.com/research/papers/liquidity-coupling.md) |
| Verification slower than decisions | Reputation-weighted auction, quorum or challenge verification | [Agent Coupling Protocol](https://docs.kalarislabs.com/research/papers/agent-coupling-protocol.md) |
| All three together | Agent Coupling Protocol (ACP) | [Orchestration Trilemma](https://docs.kalarislabs.com/research/papers/orchestration-trilemma.md) |
| Untrusted content steers the agent | Context barrier, stake before high-impact actions, reputation cascades, verifiable execution | [Economic safeguards against IDPI](https://docs.kalarislabs.com/research/papers/economic-safeguards-indirect-prompt-injection.md) |

## Headline simulation results

| Result | Value | Paper |
| - | - | - |
| Task success, barrier vs linear context pricing | 0.73 vs 0.42 | [Marginal Utility of Context](https://docs.kalarislabs.com/research/papers/marginal-utility-of-context.md) |
| Mean cascade depth, α = 0.20 vs α = 0.05 | 1.4 vs 4.7 (−70.2%) | [Liquidity Coupling](https://docs.kalarislabs.com/research/papers/liquidity-coupling.md) |
| Systemic failure rate, ACP vs four baselines | 0.011 vs 0.089–0.156 | [Orchestration Trilemma](https://docs.kalarislabs.com/research/papers/orchestration-trilemma.md) |
| Attack success, combined safeguards vs sandbox | 2.1% vs 38.4% | [Economic safeguards against IDPI](https://docs.kalarislabs.com/research/papers/economic-safeguards-indirect-prompt-injection.md) |

> [!WARNING]
>
> None of these results has been replicated with deployed agents. Each paper lists the simulator's simplifications in
> its limitations section.

## Code

The Liquidity Coupling escrow mechanism and its 10,000-node simulator are in
[KalarisLabs/liquidity-coupling-](https://github.com/KalarisLabs/liquidity-coupling-).


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