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.

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

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.

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

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

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.

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

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.

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

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.

Does the model actually pick the right tool from your MCP server?

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