4 articles
specificationadvanced9 min read
Agent MCP Server Discovery Specification
How AI agents discover, register, and connect to Model Context Protocol (MCP) servers — manifests, capability handshake, auth, and trust verification.
specificationadvanced8 min read
Agent Prompt Template Versioning Specification
Agent prompt template versioning treats prompts as semver artifacts with frozen text, parameter contracts, evaluation gates, and rollback paths so production agents can ship prompt changes safely.
specificationintermediate7 min read
Agent Tool Side-Effect Disclosure Specification
Side-effect disclosure annotates agent tools with read/write semantics, irreversibility, and risk class — letting the agent runtime branch between auto-execute, confirmation, and block.
comparisonadvanced8 min read
MCP vs Function Calling vs OpenAI Plugins: AI Agent Tool Integration Architectures Compared
MCP vs function calling vs plugins compared for AI agent tool integration: discovery scope, maintainability, and documentation patterns for 2026 stacks.