I built a small Python library to add retries, caching, fallbacks, budgets, and guardrails around native LLM SDK calls

Releases·Tool Release

Dev releases ftouter, an open-source Python package that auto-falls-back across free-tier LLM APIs to stop silent app failures

A developer released ftouter, an open-source Python package that routes app traffic across multiple free-tier LLM providers and automatically falls back to the next available model when one fails. It targets a common pain point: models on free tiers getting rate-limited or deprecated without warning, crashing apps in production. The package is early-stage and community feedback is being solicited.

A developer released ftouter, an open-source Python package that routes app traffic across multiple free-tier LLM providers and automatically falls back to the next available model when one fails. It targets a common pain point: models on free tiers getting rate-limited or deprecated without warning, crashing apps in production. The package is early-stage and community feedback is being solicited.

Releases·Tool Release

Dev releases boothpy v0.5.0-beta, an open-source Python LLM checkpoint library, and seeks developers to stress-test it

An independent developer released v0.5.0-beta of boothpy (BOOTH), an open-source Python library providing a zero-dependency, provider-agnostic checkpoint layer between LLM calls and applications (LLM → BOOTH → accept/retry/reject). The release post recruits developers to test the library across providers like OpenAI, Anthropic, Groq, Mistral, and Ollama, with a GitHub issue open for provider integration examples and benchmarks. It matters to practitioners building reliability tooling who want an early-stage, MIT-licensed option to evaluate.

An independent developer released v0.5.0-beta of boothpy (BOOTH), an open-source Python library providing a zero-dependency, provider-agnostic checkpoint layer between LLM calls and applications (LLM → BOOTH → accept/retry/reject). The release post recruits developers to test the library across providers like OpenAI, Anthropic, Groq, Mistral, and Ollama, with a GitHub issue open for provider integration examples and benchmarks. It matters to practitioners building reliability tooling who want an early-stage, MIT-licensed option to evaluate.

Products·building

Python SDK for building and deploying LLM agents with built-in infra

Custodian Labs is a Python SDK that simplifies building and deploying LLM agents by handling surrounding infrastructure. It offers model-agnostic switching, built-in RAG, and a simple deploy() method, targeting developers who want to avoid rebuilding common infrastructure for each LLM app.

Custodian Labs is a Python SDK that simplifies building and deploying LLM agents by handling surrounding infrastructure. It offers model-agnostic switching, built-in RAG, and a simple deploy() method, targeting developers who want to avoid rebuilding common infrastructure for each LLM app.

AI Trust Gateway is a lightweight, zero-cost Python proxy middleware built with FastAPI that acts as an input firewall for LLMs. It catches prompt injections and redacts sensitive API keys locally before they reach AI models, helping developers building custom AI apps secure their interactions.

Releases·Tool Release

Developer releases Custodian Labs, a Python SDK bundling LLM agent infrastructure

A developer shared Custodian Labs, a Python SDK for building and deploying LLM agents without manually wiring up surrounding infrastructure. The SDK is model-agnostic and ships with built-in RAG, multi-agent support, and privacy/PII handling. It targets practitioners rebuilding the same deployment scaffolding for every LLM app.

A developer shared Custodian Labs, a Python SDK for building and deploying LLM agents without manually wiring up surrounding infrastructure. The SDK is model-agnostic and ships with built-in RAG, multi-agent support, and privacy/PII handling. It targets practitioners rebuilding the same deployment scaffolding for every LLM app.