#llm

No repository description published.

A local meter that prices subscription-agent usage and caps API-key spend before the next dollar leaves. One binary, one local SQLite ledger, no cloud. MIT.

Pre-flight risk linting for MCP and agent tools — check the blast radius before your agent runs them.

Prior art for AI agents. Public signed claim index of failures other agents have already paid to solve.

Context.dev CLI: scrape, crawl, search and screenshot the web from your terminal. Returns LLM-ready markdown or JSON.

Make Claude Code and Codex prove their work — runs advance only when your tests, lint, and typecheck actually pass. Open-source Go runtime.

High-Precision GraphRAG. Native to PostgreSQL.

A lightweight and elegant AI Agent framework

Host + AG-UI bridge for LangGraph agents — agent-spec loading, layered config, and typed in-process streaming

Paired-scientist AI assistant for reproducible research — lightweight Go runtime, lifecycle hooks, manuscript integration, 12 baseline scientific skills. PicoClaw-compatible.

The verification layer for LLM serving — your check runs on every answer before it ships, with a signed, tamper-evident receipt for each decision and a distribution-free bound on wrong answers served. Proof over prediction.

A single interface in Go to get inference from multiple llm/ai providers using their official SDKs

Open-source financial accountability for AI-agent work. Connect Claude Code and Codex activity to cost evidence, attribution, runway, and evidence-backed next actions via CLI or MCP; customize the source-built macOS Glance. Local transcript analysis stays on your machine.

Booplex pi extensions — AI advisor councils (@booplex/bpx-consult) and model endpoint management (@booplex/bpx-endpoints) for the pi coding agent

Dual-provider LLM gateway (OpenAI + Anthropic) with an agentic tool-use loop over milpa/tool-runtime. Apache-2.0.

Sphinx shared components for git-pull projects

The agent harness: sandbox any agent, verify each step by replaying it against real state, and keep a deterministic trace.

No repository description published.

Secure-by-default .NET guardrail library for LLM apps with 3-layer protection (regex, ML, LLM-as-Judge) covering prompt injection, jailbreaks, toxicity, PII masking, and format validation.

AI-powered Git copilot for your terminal. Talk to Git in plain English, automate commits, review code, and manage repos with a beautiful TUI.

Training-free KV cache compression via E8 lattice VQ. 2-bit KV that preserves retrieval (30/30 NIAH vs TurboQuant 0/30). Calibration-free, 9 architectures validated.

Uniform streaming LLM interface for voice agents — OpenAI, Anthropic, Google, Vertex, Bedrock

Autonomous AI agents you deploy in 30 seconds across any channel. Zero-config CLI, 37 channel adapters, HEXACO personality and cognitive memory powered by AgentOS, five-tier prompt-injection defense.

Security boundary CLI proxy between AI coding agents and your shell — redacts secrets, neutralizes prompt injection, wraps untrusted terminal output, and audits every command. Single Rust binary.

Don't build another AI. Manage the ones you have. — Kubernetes for AI agents

Harness Development Kit

Modular Python voice I/O for AI applications. Text-to-speech, speech-to-text, and voice cloning with local-first defaults and remote provider support. One interface across Piper, Supertonic, OpenAI, OmniVoice, and more.

C# tool for converting files and office documents to Markdown.

Language model tokenization at GB/s

Loom is a high-performance, graph-based AI workflow engine for Go. It allows you to build complex agentic systems by defining multi-step workflows as directed graphs, seamlessly connecting them to any large language model, and persisting execution state for long-running or interrupted tasks.