核心模型
1. 行动空间质量 2. 观察质量 3. 恢复质量 4. 上下文预算质量
affaan-m/ECC
Review agent-harness-construction's use cases, installation, workflow, and original source instructions.
npx skills add https://github.com/affaan-m/ECC --skill "docs/zh-CN/skills/agent-harness-construction"Source checked Jul 28, 2026·Refresh due Oct 26, 2026
Limited to facts supported by the pinned source
当你在改进智能体的规划、调用工具、从错误中恢复以及收敛到完成状态的方式时,使用此技能。
npx skills add https://github.com/affaan-m/ECC --skill "docs/zh-CN/skills/agent-harness-construction"The pinned source contains about 26 English words and 9 usable sections. That evidence supports a source profile, not a complete guide.
26 source words · 9 usable sections
What the source actually contains
Sections are extracted automatically from the pinned SKILL.md and link back to the source.
1. 行动空间质量 2. 观察质量 3. 恢复质量 4. 上下文预算质量
1. 使用稳定、明确的工具名称。 2. 保持输入模式优先且范围狭窄。 3. 返回确定性的输出形状。 4. 除非无法隔离,否则避免使用全能型工具。
对高风险操作(部署、迁移、权限)使用微工具。 对常见的编辑/读取/搜索循环使用中等工具。 仅当往返开销是主要成本时使用宏工具。
Choose a different workflow
Design and optimize AI agent action spaces, tool definitions, and observation formatting for higher completion rates.
A separate implementation from affaan-m/ECC; compare its source, maintenance signals, and permission requirements.
Open source detailReview agent-harness-construction's use cases, installation, workflow, and original source instructions.
A separate implementation from affaan-m/ECC; compare its source, maintenance signals, and permission requirements.
Open source detailFAQ
The source record exposes this install command: npx skills add https://github.com/affaan-m/ECC --skill "docs/zh-CN/skills/agent-harness-construction". Inspect the command and pinned source before running it.
Quality breakdown
Based on traceable docs and repository signals; stars are not treated as quality.
Compare before choosing
These links are selected from shared tasks, functions, stacks, platforms, and same-name variants. Compare the source owner, documentation, permissions, and maintenance signals.
当你在改进智能体的规划、调用工具、从错误中恢复以及收敛到完成状态的方式时,使用此技能。
智能体输出质量受限于:
每个工具响应都应包括:
status: success|warning|errorsummary: 一行结果next_actions: 可执行的后续步骤artifacts: 文件路径 / ID对于每个错误路径,应包括:
跟踪: