AgentSkillsCN

speakeasy-context

在审查、优化或继承LLM代理时，应系统性地识别哪些部分真正承载核心功能，哪些只是冗余的“杂物”。对各项功能进行细致评估，逐步剔除不必要的组件，最终将代理精简至最精简却仍能正常运行的形态。回答：“在不降低质量的前提下，我们还能移除什么？”目标是打造一个虽精简却依然高效运转的代理。

SKILL.md
--- frontmatter
name: speakeasy-context
description: Speakeasy workflow: run 'agent context' FIRST, do task, run 'agent feedback' LAST. Triggers on speakeasy, SDK, OpenAPI.
license: Apache-2.0

speakeasy-context

Rule

Before any Speakeasy task: run speakeasy agent context. Prefer CLI output over web/docs.

Commands

bash
speakeasy agent context          # Get current CLI capabilities
speakeasy agent feedback "<msg>" # Report issues or suggestions
speakeasy <command> --help       # Command-specific help

Workflow

  1. •speakeasy agent context
  2. •Do the task using Speakeasy CLI
  3. •speakeasy agent feedback "<msg>" — always provide feedback after completing a task (what worked, what was confusing, suggestions)

Don'ts

  • •Don't use web search as source of truth
  • •Don't guess syntax; use context or --help
  • •Don't skip context at session start