Skip to content

AOP v0.5.0 — Strategy Semantics + MCP Schema-Derived Output Validation

Choose a tag to compare

View all tags
@joy7758 joy7758 released this 04 Mar 03:53
· 59 commits to main since this release
v0.5.0
f64565b

AOP v0.5.0 - Strategy Semantics Fixtures + MCP Interop Gates

Summary

v0.5.0 promotes phase-1 semantics to a reproducible conformance
baseline using schema-validated fixtures and CI gates, with no runtime
implementation required in this repository.

What's Included

  • Combination semantics fixtures for policy/decision strategy deltas:
    • same policy/object pair
    • strategy-specific expected decisions (first_match,
      deny_overrides)
  • Semantic mismatch guard for combo expected decisions:
    • schema-valid but semantically contradictory fixture is rejected by
      CI semantic checks
  • MCP interop schema-derived validation gate:
    • derive output schema from each tool fixture schema.outputs
    • validate expected outputs per tool prefix
    • reject invalid outputs per tool prefix
  • Published v0.5 semantics contract:
    • V0_5_SEMANTICS.md with fixture references and external anchors
      (XACML, MCP tools output schema guidance, JSON Schema
      unevaluatedProperties)

Governance

  • AEP-0004 status moved to Accepted with explicit phase-1
    acceptance criteria.

Compatibility

This is a MINOR release in v0.x. Review migration and conformance notes
if your implementation consumes new v0.5 fixture families.

Validation Baseline

  • Schemas compile under draft2020
  • Positive fixtures pass
  • Negative fixtures are rejected
  • v0.5 combo and MCP gates are enabled in CI

Short External Blurb (Copy/Paste)

中文版（10行）

  1. AOP v0.5.0 已正式发布，定位为 AI Agent 的对象契约层。
  2. 该版本以“可验证工件”为核心：schemas + fixtures + CI gates。1
  3. 它不绑定任何 runtime，实现可替换、契约可复用。
  4. combo 线展示同一 policy/object 在不同策略下的决策差异。
  5. first_match 与 deny_overrides 都有对应 expected decision 工件。2
  6. 同时包含语义负例，确保“策略标签不能乱写”可被 gate 拦截。
  7. mcp 线从 tool 的 schema.outputs 派生输出校验约束。
  8. expected 输出必须通过，invalid 输出必须被拒绝。
  9. 这与 MCP 对 output schema 的责任模型一致：server MUST、client SHOULD。3
  10. v0.5.0 是一个可引用、可复现、可裁决的 MINOR 基线。

English (10 lines)

  1. AOP v0.5.0 is now released as the object-contract layer for AI agents.
  2. The baseline is built on verifiable artifacts: schemas, fixtures, and CI gates. 1
  3. It remains runtime-agnostic: no reference runtime is required in-repo.
  4. The combo track demonstrates decision deltas for one policy/object pair.
  5. It ships expected decisions for first_match and deny_overrides. 2
  6. It also includes a semantic-invalid fixture to prevent label/behavior mismatch.
  7. The mcp track derives output validation directly from tool.schema.outputs.
  8. Expected outputs must pass; invalid outputs must fail in CI.
  9. This aligns with MCP output-schema responsibilities (server MUST, client SHOULD). 3
  10. v0.5.0 is a citeable, reproducible, and adjudicable MINOR baseline.

  1. JSON Schema 2020-12 unevaluatedProperties: json-schema.org ↩ ↩2

  2. XACML 3.0 Core (combining algorithms): docs.oasis-open.org ↩ ↩2

  3. MCP Tools spec (output schema responsibility): modelcontextprotocol.io ↩ ↩2

Assets 2
Loading