Agent 标准生态

The Agent standards ecosystem splits agent products into portable contracts. Each standard owns one layer of meaning and links to the others through stable refs instead of swallowing their responsibilities.

Agent Tool 的位置

Agent Tool owns portable tool facts: declarations, surfaces, capability refs, input/output contracts, execution profiles, permission profiles, invocations, progress, results, errors, and audit refs.

Tool explains what capability was exposed, how it was selected, what call was made, how it progressed, what result came back, and how native protocol ids relate to runtime, UI, evidence, policy, artifact, knowledge, and skills.

当前标准

Standard	Role	Site	LLM context	Repository
Agent Knowledge	Source-grounded knowledge packs for agents.	site	llms-full	repo
Agent UI	Interaction surfaces for agent products.	site	llms-full	repo
Agent Runtime	Execution facts, controls, tasks, tools, and recovery.	site	llms-full	repo
Agent Evidence	Evidence, provenance, verification, review, replay, and export.	site	llms-full	repo
Agent Policy	Risk, permission, approval, retention, waiver, access, and policy decision facts.	site	llms-full	repo
Agent Artifact	Durable deliverables, versions, parts, previews, exports, source links, and handoff packages.	site	llms-full	repo
Agent Tool	工具声明、工具面、调用、进度、结果、权限和审计引用。	site	llms-full	repo
Agent Context	Agent 当前可用、选择、预算、组装、注入、压缩、缺失与脱敏的上下文事实。	site	llms-full	repo

边界规则

Agent Knowledge -> what durable source-grounded context an agent can use
Agent Runtime   -> how agent work is accepted, executed, controlled, and resumed
Agent UI        -> how agent work is projected into user-visible surfaces
Agent Evidence  -> why an agent outcome can be trusted, reviewed, replayed, and exported
Agent Policy    -> whether an agent action may proceed and under which constraints
Agent Artifact  -> what durable deliverable the agent produced and how it changes
Agent Tool      -> what capability was exposed, invoked, progressed, and returned
Agent Context   -> Agent 当前可用、选择、组装、压缩、缺失并注入了什么上下文

No standard should become the whole stack. A compatible implementation should preserve native ids and link across standards with refs.

未来候选标准

Candidate	Why it may become a standard
Agent Evaluation	Acceptance scenarios, rubrics, eval runs, quality gates, and evidence-backed benchmark records.
Agent Workflow	Portable multi-step work plans, scene launches, background jobs, and handoff states.
Agent Model Routing	Task profiles, model candidates, routing decisions, fallback, quota, and cost records.
Agent Channel	Remote runtime, A2A peers, webhooks, deep links, browser sessions, and desktop bridges.

These candidates should remain design notes until they can be specified without relying on one product implementation.