AGENT SKILL

second-call-boundary

Post-tool LLM calls are allowed when the first model response requested server-executed tools and the second call incorporates exact tool results

License 许可
MIT
最近更新
2026-11-14
Stars
★ 6
Forks
0

01 前言 Front Matter

name
second-call-boundary
description
Post-tool LLM calls are allowed when the first model response requested server-executed tools and the second call incorporates exact tool results

03 SKILL.md 正文

Second-Call Boundary Protocol

Policy Overview

Post-tool LLM calls are allowed when the first model response requested server-executed tools and the second call incorporates exact tool results.

Prohibited Patterns

Do not add any of the following second-call wrapper variations:

  • JSON/schema repair calls
  • Fallback re-prompts
  • Spell-repair calls
  • Missing-tool retry prompts
  • Prompt-based retry calls

The only allowed semantic sidecar is the FastEmbed intent classifier.

04 同仓库更多 Skills