agentic-harness
Use when designing, reviewing, or debugging agent loops, harness construction, tool/action spaces, or multi-agent orchestration.
- License 许可
- MIT
- 最近更新
- 2026-10-08
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- agentic-harness
- license
- MIT
- metadata
- { "owner": "Harley Bartles", "status": "active", "use_when": [ "designing or reviewing an agent loop.", "choosing or constructing a harness for tools and actions.", "orchestrating multiple agents or merging parallel results." ], "source-id": "agentic-harness", "source-path": "skills/agentic-harness/SKILL.md", "related_skills": [ "agent-evaluation", "research-ops", "release-engineering" ], "do_not_use_when": [ "another more specific skill owns the task." ], "provenance-name": "Agentic Harness first-party skill", "source-category": "first_party" }
- description
- Use when designing, reviewing, or debugging agent loops, harness construction, tool/action spaces, or multi-agent orchestration.
02 资源文件 5
- 参考文档agents/openai.yaml484 B
- 资产文件assets/authority/authority.yaml942 B
- 资产文件assets/authority/CITATIONS.md1.4 KB
- 资产文件assets/authority/source-map.yaml433 B
- 参考文档references/operational-guidance.md1.9 KB
03 SKILL.md 正文
Agentic Harness
Use this skill when designing or reviewing agent loops, harness construction, tool/action spaces, and multi-agent orchestration.
When to Use
- Designing or reviewing an agent loop.
- Choosing or constructing a harness for tools and actions.
- Orchestrating multiple agents or merging parallel results.
Core Pattern
- Model the loop explicitly: input, plan, tool call, observation, output, stop check.
- Define tools with typed schemas and documented side effects; keep the action space small and composable.
- Isolate state between parallel agents (for example, git worktrees or sandboxed contexts) and merge through explicit handoff.
- Set iteration and error budgets to prevent silent infinite loops.
- Test the harness against deterministic trajectories before scaling to open-ended work.
Common Mistakes
- Leaking shared mutable state between parallel agents. → Isolate contexts and merge explicitly.
- Implicit infinite loops. → Set iteration and error budgets.
- Opaque tool failures. → Normalize errors into structured observations.
Load references/operational-guidance.md for deeper coverage of loop design, tool spaces, orchestration, and harness construction.
04 同仓库更多 Skills
writing
Use when drafting, revising, or reviewing prose for human readers and clarity, an authorised voice, or reader-fatigue concerns may interact.
writing-with-clarity
Use when drafting, revising, or reviewing prose intended for human readers, including documentation, README files, reports, explanations, UI copy, errors, comments, commit messages, and pull request text.
writing-style
Use when revising or reviewing human-facing prose against an authorised voice card or a contextual reader-fatigue profile.
skill
暂无简介
writing-skills
Use when creating new skills, editing existing skills, or verifying skills work before deployment.
writing-roadmaps
Use when a goal is too large for one writing-plans plan and requires a sequenced roadmap of consecutive plans.