git:20261008.c74aa9d to git:20261018.8261244

13 added, 11 removed. Audit A to A.

---

name: skill-invocation-router

description: Design and test explicit human, implicit model or agent, programmatic application, bounded skill-composition, and harness activation policies for an Agent Skill catalog.

license: MIT

metadata:

lesson: "25"

---

- # Skill invocation router

+ <a id="skill-invocation-router"></a>

- Use this skill when a host needs an auditable activation policy rather than one undifferentiated `invocable` flag.

+ # 技能调用路由器

- 1. Read `references/invocation-model.md` and classify the requested channel.

- 2. Review `assets/host-policy.json` as an example adapter configuration, not a portable standard.

- 3. Run `python3 scripts/simulate_invocation.py --policy assets/host-policy.json --actor ACTOR --name NAME --description DESCRIPTION --query QUERY [--explicit-name NAME] [--caller-name NAME] [--depth N] [--user-invocable true|false] [--disable-model-invocation true|false]`.

- 4. For a human, application, skill, or harness request, require an exact discovered name and its channel-specific allowlist.

- 5. For a skill caller, also require caller identity, a non-cyclic target, and a bounded composition depth.

- 6. For a model or autonomous agent request, remove candidates that the actor or recognized host extensions make ineligible.

- 7. Score only the remaining descriptions. Select the strongest eligible match or abstain when no eligible candidate clears the threshold.

- 8. Return the JSON decision with the adapter, channel, score, and policy reason.

+ 当宿主需要可审计的激活策略，而不是一个不加区分的 `invocable` 标记时，使用此技能。

- Activation loads instructions. It does not approve tools, filesystem changes, network access, secret use, or bundled scripts.

+ 1. 阅读 `references/invocation-model.md`，对请求渠道分类。

+ 2. 将 `assets/host-policy.json` 视为适配器配置示例进行审查，不要将其当作可移植标准。

+ 3. 运行 `python3 scripts/simulate_invocation.py --policy assets/host-policy.json --actor ACTOR --name NAME --description DESCRIPTION --query QUERY [--explicit-name NAME] [--caller-name NAME] [--depth N] [--user-invocable true|false] [--disable-model-invocation true|false]`。

+ 4. 对人工、应用、技能或运行框架请求，要求使用发现到的准确名称，以及相应渠道的允许列表。

+ 5. 对技能调用方，还要求调用方身份、无循环的目标，以及有界的组合深度。

+ 6. 对模型或自主智能体请求，先移除按执行者或已识别宿主扩展规则判定为不合格的候选项。

+ 7. 仅对剩余描述评分。选择最匹配的合格候选项；若没有合格候选项超过阈值，则不选择。

+ 8. 返回 JSON 决策，包含适配器、渠道、评分和策略原因。

+ 激活只会加载指令，并不批准工具调用、文件系统修改、网络访问、秘密使用或包内脚本执行。