# Agent Execution Knowledge Sources: What AI Agents Need to Look Up
Status: public
Confidence: medium (0.83) (verified)
Last verified: 2026-06-02
Generation: ai_structured


## TL;DR

Agent-ready content should prioritize the materials an agent needs to select tools, retrieve context, call APIs, inspect code, verify outcomes, and avoid unsupported claims.

## Core Explanation

For AnchorFact, the practical corpus map is sevenfold: official docs and API references, code repository knowledge, tool and integration docs, citation-ready technical facts, troubleshooting and version compatibility, security and governance boundaries, and real workflow/web task context. These map directly onto agent tool surfaces, MCP-connected systems, and benchmark environments that test web, API, shell, database, code, and multi-step decision tasks.

## Source-Mapped Facts

- OpenAI Agents SDK documentation says tools let an agent fetch data, call external APIs, execute code, or use a computer. ([source](https://openai.github.io/openai-agents-js/guides/tools/))
- Model Context Protocol documentation describes MCP as an open-source standard for connecting AI applications to external systems, including data sources, tools, and workflows. ([source](https://modelcontextprotocol.io/docs/getting-started/intro))
- AgentBench presents a benchmark with eight distinct environments for evaluating LLM agents' reasoning and decision-making abilities. ([source](https://arxiv.org/abs/2308.03688))
- WebArena builds a realistic web-agent environment around functional websites in e-commerce, forum, collaborative software development, and content management domains. ([source](https://arxiv.org/abs/2307.13854))

## Further Reading

- [OpenAI Agents SDK Tools](https://openai.github.io/openai-agents-js/guides/tools/)
- [Model Context Protocol Introduction](https://modelcontextprotocol.io/docs/getting-started/intro)
- [AgentBench: Evaluating LLMs as Agents](https://arxiv.org/abs/2308.03688)
- [WebArena: A Realistic Web Environment for Building Autonomous Agents](https://arxiv.org/abs/2307.13854)