tool-selection

Context engineering for AI agents. ~80% fewer tokens. Fix tool overload. Skills and memory with in-process BM25 and semantic retrieval. Progressive Disclosure. No vector DB.

Dynamically expose tools from proxied servers based on an Agent Persona

🧰 Tool Search Engine for Efficient Tool Use in AI Agents

Explore Functions, Tools and Agents with LangChain along with LangChain Expression Language

DeepSeek Harness MCP tool search for large catalogs: 1,000 MCP tools behind 2 MCP-facing schemas, exact-schema calls, allow/deny controls, and a local calculator.

Paper Implementation: Semantic Tool Discovery for Large Language Models: A Vector-Based Approach to MCP Tool Selection https://arxiv.org/html/2603.20313v1

Local-first MCP context budget and tool-selection verifier for agentic coding environments.

In-process Go library for intent-aware MCP tool selection: keyword intent detection, priority-ordered selection rules, token-budget pruning, and a per-tool enrichment pipeline.

A new package that helps users compare and choose the right data analysis tool by providing structured, expert-level insights. Users input their specific data analysis needs, project requirements, or

A tool that lets you run many tools at the same time with Ai for bug bounty !!

The selection layer for agents: which tool an agent can use, should use, and pays — between discovery (MCP/ARD) and settlement (x402/AP2)

Skill for choosing image generation, diagram, screenshot, and README visual asset tools

Open-source RAG router for MCP — fronts all your MCP servers and exposes only the relevant tools per query, killing context overload.

🤖 Minimal CLI agent using single-turn ReAct pattern with OpenAI LLM wrapper. Intelligently selects between calculator, string operations, and direct responses. Extensible tool system with comprehensive testing.

Contract-aware evaluation harness for diagnosing invalid tool selection in AI systems.

Least-context tool palettes for AI agents and MCP tool catalogs.

Cascading tool selector for LLM agents — narrows 1000+ tools to the best match in ~450ms using 3 progressive filters: semantic embeddings → cross-encoder reranker → micro-LLM. Works with OpenAI, Anthropic and Gemini.

Offline, zero-dependency MCP tool-router — funnel N tools to a ranked shortlist so your agent loads ~8, not 200. No embedding model, no network.

Shrink massive MCP catalogs to two tools, letting DeepSeek Harness search and call 1,000+ remote APIs efficiently.