cckyros/goal-acceptance--packages-goal-acceptance ↗★ 1
@cckyros/goal-acceptance
模型提供商无关的目标验收标准与验证执行插件，支持管理不可变验收标准、提供模型工具并拦截会话结束以引导未完工作。
AI 分析
核心用途是强制执行任务的验收标准，确保 Agent 交付质量。适合需要对 Agent 任务结果进行结构化总结和严格合规性校验的用户。基于 Cordis 框架构建。
安裝
此插件尚未提供可驗證的 bundle，或相容性檢查未通過。請先閱讀倉庫說明。 閱讀完整 README ↗
說明文件
閱讀完整 README ↗@cckyros/goal-acceptance
English | 中文
Model-provider-neutral goal acceptance criteria and validation enforcement plugin for the DeepSeek Harness (goal-acceptance).
This plugin manages immutable acceptance criteria per Goal session, provides model-facing tools (set_acceptance_criteria, validate_criterion, get_acceptance_criteria), and intercepts agent/turn-stopping to steer uncompleted work or report a final structured summary.
It is a thin Cordis wrapper over @cckyros/goal-acceptance-core, which is a framework-agnostic npm package that can also be used in OpenClaw, Cursor, Claude Code, or other agent runtimes.
For the portable Agent Plugin and MCP packaging, see @cckyros/goal-acceptance-mcp.
Plugin
apply(ctx) provides ctx.goalAcceptance and registers model-facing tools and prompts.
Model Experience
Criteria Confirmation
What the model sees
A structured list of required criteria, verification methods, and capability limitations before autonomous execution begins.
Token effect
Criteria stay in the session log and model context until the Goal concludes.
KV Cache effect
Append-only; new criterion validations follow the session event stream.
Known Limitations and Deferred Work
- Interactive criteria modification after confirmation is deferred; subsequent modifications start a new Goal session.