agent-retry

---

The Vedic Principle

In the Bhagavad Gita, Lord Krishna reveals to Arjuna: "na jāyate mriyate vā kadācin" — the soul is never born, never dies. It returns, again and again, until liberation.

punarjanma brings this eternal principle to LLM agent engineering. When your API call fails, when the model returns garbage, when the network drops — the agent returns. Exponential backoff is not failure. It is the dharmic cycle of persistence, each attempt purified by the lessons of the last.

Just as the soul accumulates wisdom across lifetimes to finally achieve moksha, your agent accumulates context across retries to finally achieve its goal. The dead letter queue is not defeat — it is the end of a cycle, the completion of karma.

---

How It Works

flowchart LR
    A[LLM Call] --> B{Success?}
    B -- Yes --> C[Return Result]
    B -- No --> D[punarjanma]
    D --> E{Attempts < Max?}
    E -- Yes --> F[Exponential Backoff]
    F --> A
    E -- No --> G[Dead Letter Queue]
    style D fill:#6b21a8,color:#fff

---

Installation

pip install punarjanma

Or from source:

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

Quick Start

from punarjanma import *

# See examples/ for full usage

---

The Vedic Arsenal

punarjanma 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+.

---

Contributing

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

---