🌊 मर्यादा

maryada

Ramayana — Rama's dharma

Sacred Boundaries — Rama Maryada Purushottam

Concurrency limiter for LLM agents. Max parallel calls, queue depth, backpressure.

Formerly agent-limiter — Part of the Vedic Arsenal: 100 production-grade Python libraries for LLM agents, each named from the Vedas, Puranas, and Mahakavyas.

Ram is called Maryada Purushottam — the supreme upholder of sacred limits. He never exceeded his dharmic boundaries, even when it would have been easier to do so.

maryada brings this principle to LLM rate limiting. The sacred boundary — tokens per minute, requests per hour, cost per day — is not a constraint. It is dharma. Exceeding it is adharma, causing harm to the entire system.

The token bucket, the sliding window, the ModelLimits presets — these are Ram's maryada, drawn around your API calls to protect not just your system, but the shared infrastructure that all agents depend upon.

flowchart LR
    A[API Call] --> B[maryada limiter]
    B --> C{Tokens Available?}
    C -- Yes --> D[Allow + Consume]
    C -- No --> E{Wait or Reject?}
    E -- Wait --> F[Sleep + Retry]
    E -- Reject --> G[429 Error]
    style B fill:#3b82f6,color:#fff

pip install maryada

Or from source:

git clone https://github.com/darshjme/maryada.git
cd maryada && pip install -e .

from maryada import *

# See examples/ for full usage

maryada is one of 100 libraries in darshjme/arsenal — each named from sacred Indian literature:

Each library solves one problem. Zero external dependencies. Pure Python 3.8+.

1. Fork the repo
2. Create feature branch (git checkout -b fix/your-fix)
3. Add tests — zero dependencies only
4. Open a PR