Harness Design Patterns
v1.1.0Agent harness 架构设计模式知识库。当设计多阶段任务的上下文传递、选择多 agent 协调方式、或规划 hook 系统架构时使用。纯设计指南，无可执行代码。不用于安装 hook 脚本（用 agent-hooks）或运维监控（用 agent-ops）。
⭐ 0· 95·0 current·0 all-time
by_silhouette@lanyasheng
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description match the included content: a set of harness design patterns and examples. The skill declares no required binaries, env vars, or installs, which is appropriate for a documentation-only skill. Note: the docs reference environment variables (HARNESS_PROFILE, HARNESS_DISABLED_HOOKS, TMPDIR), shell utilities (bash, jq, grep, mv), and specific filesystem locations (sessions/*, ~/.openclaw/shared-context). Those references are reasonable for implementation guidance but are not declared as requirements — verify your runtime will provide those tools/paths if you adopt the patterns.
Instruction Scope
SKILL.md is purely prose and examples; there are no executable files. However it includes concrete shell snippets and hook configurations that, if implemented, instruct agents or hooks to read/write local files (handoff/denials/bracket files), inspect transcripts, and inject prompts via hooks. This behavior is consistent with the skill's purpose (harness design) but means following the doc will create agents that read/write session state on disk and run hook scripts — review any concrete implementation for safety and correct paths.
Install Mechanism
No install spec and no code files — lowest-risk pattern for a skill. Nothing is downloaded or written by the skill itself.
Credentials
The skill requests no credentials and declares no required env vars, which is appropriate for a patterns knowledgebase. Still, the documentation references several env vars and filesystem locations used to control behavior (e.g., HARNESS_PROFILE, HARNESS_DISABLED_HOOKS, TMPDIR, sessions/*). These are not secrets, but you should be aware the recommended patterns assume access to local filesystem and typical shell utilities.
Persistence & Privilege
always:false and user-invocable:true. The skill does not request persistent presence or special platform privileges and does not modify other skills' configs. Autonomous model invocation default is unchanged and is appropriate for a documentation skill.
Assessment
This skill is a documentation-only knowledgebase about harness design — it doesn't ask for credentials or install code itself. Before adopting its patterns: 1) recognize the guide instructs creating hooks and shell scripts that will read/write session files (sessions/*, TMPDIR, ~/.openclaw/shared-context) and use common shell tools (bash, jq, grep); ensure those paths/tools are acceptable in your environment. 2) The SKILL.md references env vars (HARNESS_PROFILE, HARNESS_DISABLED_HOOKS) — these are not secrets but are expected to be set by the implementer. 3) If you implement the examples, review any hook scripts for correct file paths, locking behavior, and to avoid accidentally exposing sensitive files; test in a sandbox before using in production. 4) The doc cites internal Claude Code behaviors and external links that may change; validate assumptions against your agent runtime. Overall the package is coherent and appropriate as a design reference, but treat code snippets as templates to review rather than ready-to-run artifacts.Like a lobster shell, security has layers — review code before you run it.
latestvk9783m7yykvxb389r6fcdacr9s84bnmj
License
MIT-0
Free to use, modify, and redistribute. No attribution required.