/install hizal-search
Hizal Search
Search before you build. Don't rediscover what the team already decided.
Semantic Search
Start with 2-3 broad searches using different phrasings:
hizal__search_context(query="\x3Ckey concept from the task>")
hizal__search_context(query="\x3Cticket id or feature name>")
hizal__search_context(query="\x3Crelated subsystem or endpoint>")
Narrow by Scope
# Project-specific knowledge and conventions
hizal__search_context(query="\x3Cconcept>", scope="PROJECT", project_id="\x3Cid>")
# Prior agent memory / investigation notes
hizal__search_context(query="\x3Cconcept>", scope="AGENT", chunk_type="MEMORY")
# Org-wide principles and standards
hizal__search_context(query="\x3Cconcept>", scope="ORG")
Filter by Chunk Type
hizal__search_context(query="\x3Cconcept>", chunk_type="KNOWLEDGE")
hizal__search_context(query="\x3Cconcept>", chunk_type="CONVENTION")
hizal__search_context(query="\x3Cconcept>", chunk_type="MEMORY")
Exact Lookup by query_key
When you know the exact key:
hizal__read_context(query_key="\x3Cexact-query-key>", project_id="\x3Cid>")
Read a Specific Chunk
hizal__read_context(id="\x3Cchunk-uuid>")
Synthesize Related Context
Pull related chunks for synthesis (read-only, never delete source chunks):
hizal__compact_context(query="\x3Cconcept>")
hizal__compact_context(query="\x3Cconcept>", scope="PROJECT", project_id="\x3Cid>")
Version History
Inspect how a chunk evolved:
hizal__get_context_versions(id="\x3Cchunk-uuid>")
Search Strategy
- Broad first — 2-3 searches with different phrasings
- Read the results — chunks contain architecture decisions, conventions, and prior work
- Narrow by scope if broad search returns too much noise
- Use query_key if you already know the key
- If an AGENT memory chunk is broadly useful — promote it with
write_knowledgeorwrite_convention
See hizal-write skill for writing back to Hizal.
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install hizal-search - 安装完成后,直接呼叫该 Skill 的名称或使用
/hizal-search触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
Hizal Search 是什么?
Retrieve existing context before building anything. Self-triggering — fires whenever the agent is about to search the codebase, recall prior decisions, look... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 53 次。
如何安装 Hizal Search?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install hizal-search」即可一键安装,无需额外配置。
Hizal Search 是免费的吗?
是的,Hizal Search 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Hizal Search 支持哪些平台?
Hizal Search 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Hizal Search?
由 Parker(@parkerscobey)开发并维护,当前版本 v0.1.0。