Python agents

Operational safety controls for autonomous AI agents - kill switch, heartbeat, circuit breaker

Cross-session memory for coding agents: plain files, git-native, no server

Boot-time integrity check for AI agent memory & convention files — catch silent convention loss, injection dilution, broken rollback anchors.

Four small, dependency-free memory and safety patterns for LLM agents: scratchpad state, fuzzy recall, pre-compaction checkpointing, and permission boundaries.

Multi-step agent safety compliance benchmarking — finds the critical depth D* where LLM agents fail to maintain safety policies across sequential interactions.

Deterministic ACID transaction and rollback engine for AI agents in <1µs

Persistent work state and minimal context for long-running AI agents

An mcp

A lightweight, in-process circuit breaker for LLM token and cost budgets. Part of the agentbelt suite.

Shared fingerprinting and storage primitives for the agentbelt suite.

Detect repeat, cyclic, and stagnation loops in AI agent tool-call sequences, with cross-session pattern learning. Part of the agentbelt suite.

Agent-aware span sampling: keeps errors, cost/latency outliers, retries, and novel calls; samples the rest. Part of the agentbelt suite.

Budget circuit breaking, loop detection, and span sampling for AI agent workflows, in one install.

Neutral scientific contracts connecting AI-native CAE products.

Verifiable, attenuable delegation for AI agents - org-rooted and verified offline

Runtime AgentCreds agent-identity enforcement for tool calls - MCP middleware and agent-to-agent (A2A) verification

A local change-intelligence layer for coding agents: know what your code change will break before you make it.

Standalone brain + knowledge-base kernel (agenti ecosystem)

Evidence-backed engineering controls for AI coding agents.

Explore and compare flight options with AI agents

Explore and compare flight options with AI agents

Multi-agent audit CLI: cost governance, decision audit, evidence chain, shadow agent detection, compliance

One AI context, every agent. One source of truth for instructions, rules, skills, agents, commands and MCP config, mirrored into every AI coding tool, with per-tool overlays for what genuinely differs.

Flags AI agent code that persists untrusted tool/external/user content into long-term memory with no sanitization or provenance marker (agent memory poisoning).

Typed sync and async Python client for the Agent Messaging Service API.

A robust, enterprise-grade, drop-in security & safety guardrail layer for AI agents and LLMs.

MCP server: URL to screenshot, PDF, or structured extract for AI agents. Free key: 150 renders/month, no card; then Starter $17/month.

opsen as an agentsh secure-sandbox adapter — machine, model and tool spend on one bill.

AgentShip CLI — the `agentship` command that loads a YAML spec, builds the agent, and runs (or streams) one turn.

AgentShip kernel — spec, context, runtime, registry, errors, middleware, and the echo engine. The thin, vendor-free harness core.

AgentShip LangGraph engine — the default real-model runtime, with a LiteLLM model source that routes one `model:` string to any provider.

AgentShip observability — the OpenTelemetry implementation of the Observer port: span-tree exporter pipeline (Phoenix/Langfuse/LangSmith), the one-per-process LiteLLM cost/token callback, and the studio launcher. The vendor-free contract lives in agentship-core.

AgentShip — an open, recipe-first harness that wires best-of-breed agent libraries behind small, stable seams. Meta-package: one install for the everyday stack (kernel + default engine + CLI).

AgentShip runtime service — the REST/SSE/WS surface over an agent, with pluggable auth adapters and tenant isolation. The irreducible in-app core a gateway cannot replace.

Identity, marketplace and messaging for AI agents in one API. Register with one call, no human needed. Hire other agents or sell your skills; pay and get paid wallet-to-wallet in USDC on Base (no custody, sealed deliveries).

Runtime control for AI agents - replay any run, prove what happened, veto what must not happen.

A lightweight safety, approval, audit, replay, and rollback layer for AI agent tools

Groq LLM client for AI-Parrot (parrot.clients.groq)

vLLM (OpenAI-compatible) LLM client for AI-Parrot (parrot.clients.vllm)

AI-Commander: a ralph-loop AI agent giving LLMs shell access via function calling (TUI + CLI)

Alibaba Cloud AgenticBAS (20260630) SDK Library for Python

LangChain integration for Anlyon: actions as tools, and chains and agents traced as Anlyon runs

LlamaIndex integration for Anlyon: actions as tools, plus a vector store on the memory namespace

Read the public Arne Kellmann consulting services API

MCP gateway for AI agents to Aster DEX public data: futures and spot market overviews, order books, klines, funding and TradFi-perp screeners, vault deposit flows, public account views. Read-only, keyless.

Athena LLM client — one seam across frontier, hosted-open (with failover), and local backends

Official Blockquote SDK and CLI for blockquote.io: Python API client for AI citability scans and report comparisons.

Host-neutral adversarial white-box and black-box auditing agent

Plug-and-play topological memory substrate for LLMs, LangChain, and autonomous agent swarms.

Find out which open-source LLMs can run on your hardware.

Provider-neutral compute transfer client for Central Storage Platform

Find the parts of your Claude Code setup that cost context on every session and never do any work.

Renamed to claudexer -- see https://pypi.org/project/claudexer/

OpenAI Agents SDK 的腾讯云 CLS Agent 可观测接入 —— 一行初始化，无需手动埋点

Unified LLM client for every major provider - one API, spec-driven wire format, real cost tracking.

Conduit — lightweight AI driver for OpenAI-compatible, Ollama, Anthropic, Gemini (Python)

Python client for the Auggie v2 SDK / RPC protocol

Explore Python coverage and replay execution with a code flashlight.

CrewAI tools for Webz.io News Search - global news in natural language with semantic ranking and rich filters

A lightweight, production-grade LLM agent toolkit in Python: streaming, MCP, lazy skills, memory, and sandboxing.

One API. Every AI. Pay per call. Python client for DexL Agents.

Persistent ACP agent harness with pluggable state

The s3:// storage backend for dirigent, compatible with any S3 API.

DocsGPT backend: chat with your documents, agents, and tools.

Durable, self-healing multi-provider LLM gateway and circuit breaker with autonomous failover, tool idempotency, and context compaction for AI agents (Claude Code, Hermes, OpenClaw, OpenCode).

Inference library for running local LLMs on modern consumer GPUs

Linux runtime bridge between FastLLM WebUI and Pi Agent

A lightweight terminal AI chatbot connected to Gemini via Vercel serverless backend.

Turn a prompt + a field schema into validated, typed JSON (Instructor over Gemini 2.5 Flash on Vertex AI).

Google-OAuth-gated LLM gateway: verify a Google ID token, then run a Gemini (Vertex AI) completion for the verified caller.

Turn a prompt + a field schema into validated, typed JSON (Instructor over Gemini 2.5 Flash on Vertex AI).

Turn a prompt + a field schema into validated, typed JSON (Instructor over Gemini 2.5 Flash on Vertex AI).

RL-Split decision plane: masked PPO/TD3/bandit agents over the decision-parameter contract

An MCP server that gives any LLM host grounded, cited retrieval over your own documents — hybrid BM25 + dense retrieval, cross-encoder reranking, and built-in eval.

Provider adapters for HarborRAG.

Core contracts, domain, ports, execution, observability, and security for HarborRAG.

MCP facade and audited tools for HarborRAG.

Haystack tool integration for Algenta: a governed-execution-aware wrapper around Haystack's own MCPToolset, plus an Agent hook pair mapping the governed-execution receipt onto Haystack's real human-in-the-loop and hook primitives

MCP gateway for AI agents to Hyperliquid public data: perp/spot market overviews, quotes, order books, candles, trades, funding, account risk and HyperEVM transfers. Read-only, keyless.

An extensible ReAct judge with native model tool calling.

LangChain / LangGraph tool integration for Algenta: a governed-execution-aware toolset wrapping a self-hosted Algenta engine's MCP tool surface

LangChain tools for DNS Doctor: scan, diagnose and fix a domain's SPF, DMARC and DKIM, check global DNS propagation, audit SPF includes. Deterministic verdicts and copy-paste records from dnsdoctor.dev, never a guessed record.

Run LangChain deepagents on opsen — machine, model and tool spend on one bill.

Human-in-the-Loop integration between LangGraph and KirokuForms

Debug LangGraph-based agents visually - live execution graph, node inspection, and run replay, all from a single line.

LiteLLM MCP Gateway config generation for a self-hosted Algenta engine: profile-aware allowed_tools/allowed_params templates, a config linter, and a real-proxy conformance test suite

LlamaIndex tool integration for Algenta: a governed-execution-aware FunctionTool list wrapping a self-hosted Algenta engine's MCP tool surface

Local NVIDIA LLM experimentation workbench

Language-agnostic LLM cost metering via HTTP proxy — track spend per customer and feature

Capture LLM traffic through a drop-in proxy, curate it into a dataset, and distill a small model that replaces the frontier one.

Standardized methodology and dataset for evaluating LLM-based digital forensic timeline analysis

FSM syntactic engine to repair truncated/broken JSON from LLMs (streaming, Pydantic, zero-regex core)

Evidence-based reliability analyzer and root-cause diagnosis engine for LLM, RAG, and Agent systems.

LLM wiki to turn multiple agentic tool sessions into knowledge base for agents and humans

Reserved for the Magnetron project.

Pygym agent for Mario Bros

MaskFlow for LangChain: a reversible PII anonymizer/deanonymizer pair (drop-in for langchain-experimental's Presidio anonymizer) plus a leak-guard callback. Indian identifiers included.

MaskFlow for LlamaIndex: a PII node postprocessor (drop-in for PIINodePostprocessor), an ingestion transform that keeps PII out of the vector store, and a reversible query engine wrapper. Indian identifiers included.