Python agents

Pion prompt cache for mlx-lm on Apple Silicon — reuse a prefix's K/V across processes and restarts

Sprag speech and language models for Pipecat: Rhythm STT, Symphony LLM, and Chorus TTS

Local-first Hybrid GraphRAG retrieval engine and MCP server for linked knowledge bases

One interface to many LLM providers and models: generation, structured output, embeddings, costs, retries and rate limits.

Python client/CLI and stdio MCP server for AI agents on the experimental PoA network.

Google Gemini plugin for PolyLLM.

Ollama (local OpenAI-compatible) plugin for PolyLLM.

A modular, multi-paradigm Retrieval-Augmented Generation (RAG) framework supporting Vector, Agentic, and Graph RAG architectures.

Shared object-storage providers for the potluckdb SDK

A security-first, zero-dependency agent runtime with an execution-based evaluation harness.

Pragmatic double machine learning with AI-learned representations: Riesz regression, star aggregation and omitted-information sensitivity analysis

Audit a repository against The Pragmatic Programmer, and give agents a practical standard to code by.

Praxis — a trusted-local workspace agent runtime for files, code, data, and private knowledge.

Praxis — a trusted-local workspace agent runtime for files, code, data, and private knowledge.

Put your engineering best practices into practice, across every AI coding agent.

Local-first GraphRAG system exposed as an MCP server

Google Drive 파일·메일 발송 MCP — ProcessGPT 딥 에이전트 작업공간(/workspace) 경로 기준

Prometheus collector for RelStorage metrics

Run Claude Managed Agents self-hosted environment workers on Promigence sandboxes

Run mini-swe-agent (SWE-bench) instances on Promigence sandboxes

Promigence sandbox client for the OpenAI Agents SDK

Evidence-backed RAG with verified citations and refusal on insufficient evidence.

A LangChain middleware that checks each tool call of an agent against the permission a person signed, and records it with Provared. Evidence, not a verdict.

ProveLLM MCP Gateway — every MCP tool call checked against an agent passport, and signed into a receipt.

Verify ProveLLM agent credentials and signed receipts — offline, with the issuer's public key only.

Haystack component that fetches web pages through Proxmint residential, datacenter and mobile proxies, with country, city and ASN targeting and sticky sessions.

Anthropic infrastructure FinOps SDK — scan, analyze, and export your Anthropic org resources

OpenAI infrastructure FinOps SDK — scan, analyze, and export your OpenAI org resources

Official Pura LLM OAuth and delegated inference SDK

py3status module showing Claude Code subscription usage

Evidence-based debugging and root-cause analysis for AI agents and LLM applications

Placeholder for Python bindings to agentplane, a durable, replayable agent runtime.

Python client SDK for PYB Agent — engine launcher included (auto-discover or spawn `pyb web`)

Rewrite Unity Catalog references in Genie Agent (Genie space) definitions; a PyDABs mutator for Declarative Automation Bundles

ACGS rule-based tool-call governance for Pydantic AI agents, as a capability

Compose a Pydantic AI agent and its features into a FastAPI application

ContextFetch web search and page reading tools for Pydantic AI

Conversation persistence for pydantic-ai-app-factory, backed by pydantic-ai-harness StepPersistence

Google Firestore MemoryStore for the PydanticAI harness Memory capability — CAS versions via transactions and idempotent operation receipts

Pydantic AI workspaces in Lizard Sandboxes.

Native Pydantic AI capability for signed, verifiable consequential agent action evidence

Per-request model selection, a feature for pydantic-ai-app-factory

Pydantic AI models served through an Otari gateway

A Pydantic AI capability that checks each tool call of an agent against the permission a person signed, and records it with Provared. Evidence, not a verdict.

Scan an agent repository for reliability and safety weaknesses from a Pydantic AI agent.

OpenAI for PyFlowRED's AI nodes - and anything that speaks the OpenAI API.

Static scanner for injected instructions and retrieval-manipulation in RAG source documents.

Agentic retrieval at scale: an agent ranks and reads passages, a chain of sub-questions names the evidence the question leaves unnamed, and one call answers with exact citations.

A minimalist, non-opinionated, event-driven agent loop with cooperative interruption.

Make an external agent (Claude, Claude Code, any CLI/HTTP agent) the partner of a Pyunto exchange diary

Fast synchronous key-value storage for pyxpo apps (react-native-mmkv shape, AsyncStorage method names).

Evidence-bound evaluation, verification, validation, and release assurance for agentic systems

Sistema multiagente que genera y valida pruebas pytest para código Python 3.10+

Lightweight tool execution tracing SDK for evaluating AI agents on QAgent.

Python SDK for the Lime Go Agentic public CRM API (JSON:API, OAuth2 client credentials).

Run your Qmpass automated tests on your own machines, securely connected to your Jira projects.

A lightweight wrapper for querying an EC2-hosted LLM via OpenAI API

Optional Python client SDK for witty-rag / rag_core REST APIs

Lightweight adversarial hallucination auditor and citation grounding engine for LangGraph and production RAG pipelines.

RAGAS with a memory and a diff view. Evaluate RAG pipelines, persist runs, and fail CI on regressions.

A lightweight Python wrapper for evaluating RAG pipelines using Ragas.

Measure whether a retrieval system is actually retrieving the right things

Private, local document search: PDFs, Office files and images, with page-level citations.

RAG Security Protocol — where security tools hook into a RAG pipeline, and what they return

The project for RAG example workflow

A domain-retargetable, open-core engine for hybrid retrieval + knowledge-graph RAG: parsing, chunking, embedding, hybrid search, reranking, graph extraction, entity resolution, ontology-driven knowledge, and a capability runtime.

MCP server for ragdesk - list KBs, ingest text, and ask your local RAG API

RAGE Data Science & ML Toolkit for fast cleaning, EDA, preprocessing, and evaluation.

Load and chunk documents, create embeddings, and retrieve context for RAG applications

A small, provider-neutral agent loop over ragleap-tools Tool objects: a step cap, a taint rule that forces approval for outbound actions after untrusted content, pause/resume approval gates, and a per-run transcript. No database. Bring your own model as a plain callable, or use the included standard-library adapter for OpenAI-compatible chat APIs. BYOK, no vendor lock-in.

Ansible roles that prepare and harden a fresh Ubuntu server: user, sudo, updates, SSH and firewall

MCP client that turns allowlisted tools on remote MCP servers into ragleap-tools Tool objects. Streamable HTTP only, standard library plus ragleap-tools, owner-configured allowlist, pinned and bounded network access. BYOK, no vendor lock-in.

Durable, tenant-scoped long-term memory for LLM applications: typed entries with trust labels, SQLite full-text search, supersession of changed facts, per-scope export and erase, and an optional tamper-evident audit log. Standard library plus ragleap-security. Single-tenant is one fixed scope. Not independently audited.

Small, auditable security primitives for the ragleap libraries: immutable tenant scopes, a tamper-evident hash-chained audit log, secret handling and redaction, and hard input limits. Standard library only. Not independently audited.

Terraform module that creates a kind cluster and installs the RagLeap Helm charts, tested on a real cluster

Pre-alpha, open, local speech-to-text and text-to-speech: licence-checked manifest, hash-verified downloads, transcribe() and synthesize() with a Whisper engine (optional extra) and an Indic Parler-TTS engine (own environment). Every model is unverified.

A framework for building, evaluating, and diagnosing RAG systems.

A local failure debugger for RAG applications: isolates retrieval, ranking, generation, and abstention failures.

Python SDK for the RAGnarok HTTP API (/api/v1): auth, streaming queries, ingest, graph and sources.

Domain-agnostic RAG toolkit: embeddings, chunking, vector stores, retrieval, ingestion, evaluation, and jobs.

ModernBERT-based RAG evaluation model

Pre-ingestion diagnostic for RAG knowledge bases, grounded in the Garani 2026 taxonomy of 33 RAG failure modes

Evaluation-first RAG pipeline: retrieval, generation, deterministic metrics, and guardrails for sensitive domains.

Evidence-first RAG retrieval diagnostics, built to run locally.

Run bigger models on smaller machines — RAM-adaptive, GPU-aware LLM inference manager

Generate evidence-based project READMEs from your terminal or browser.

13 rules on a Salesforce Agentforce topic file, before the deploy and before Article 50 (bundled Node.js runtime)

One searchable memory of your Claude Code, Codex and Cursor chats, exposed to every agent over MCP.

Typed, correctable long-term memory for Pydantic AI agents, backed by Recall by Polign

Administrative and read API clients for Redis Lens

Python client for the Reflex sidecar (`reflex serve`).

LangGraph multi-agent apps with a FastAPI backend and a Reflex UI: streaming chat, human approval, and long-running jobs

Reglow AI-Native 生成能力（Schema DSL + 生成引擎 + Agent 运行时 + 评测集）—— 依赖 reglow 基础库的附加产品包

Pre-open repo agent-surface audit: per-harness OPEN / OPEN-QUARANTINED / DO-NOT-OPEN verdict on a received repo's complete agent-control surface, before your agent's first git call.

Turn an agent security failure into a reproducible regression test.

RUMI-8M small decoder-only Transformer language model

Haystack integration for Runcrate: 200+ open models behind one OpenAI-compatible API

Bounded native Rust key value cache and attention primitives for Python LLM pipelines

vLLM plugin that serves sabi-noema-mla (shared pre-RoPE rotary MLA, rank 512, on the 16 full-attention layers of a Qwen3.5/3.8 hybrid)

An MCP which adds terminal capabilities to the agent.