Knowhere Skill#

This ClawHub skill depends on the Knowhere OpenClaw plugin and teaches agents how to use the knowhere_* tools well.

Prerequisite#

Install the plugin before using this skill:

bash

openclaw plugins install @ontos-ai/knowhere-claw

Then enable the knowhere plugin entry in OpenClaw and restart OpenClaw.

When to Use Knowhere#

Use Knowhere when the user wants to:

• ingest a local file or URL into the current scope
• inspect, summarize, or quote previously ingested documents
• inspect ingest jobs or import a completed Knowhere job
• preview, list, remove, or clear stored documents
• understand what fields exist inside the stored result package

Do not assume an uploaded attachment was already ingested. If the user asks you to use an attached file and no existing Knowhere result already covers it, call knowhere_ingest_document.

Tool Selection#

• knowhere_ingest_document for new local files or URLs
• knowhere_list_documents to find candidate docId values in the current scope
• knowhere_preview_document for a quick structural overview
• knowhere_grep for text search across chunk fields
• knowhere_read_result_file for manifest.json, hierarchy.json, kb.csv, table HTML, or other text-like files under result/
• knowhere_list_jobs, knowhere_get_job_status, and knowhere_import_completed_job for async jobs
• knowhere_remove_document and knowhere_clear_scope for cleanup

After ingesting a document, use the returned identifiers for follow-up operations instead of guessing names.

Recommended Workflow#

1. Ingest or import the document if it is not already in the store.
2. Call knowhere_list_documents if you need to confirm the right docId.
3. Call knowhere_preview_document to get a structural overview.
4. Call knowhere_grep with conditions: [{ pattern: "your query" }] for the default text search path.
5. Narrow by chunk.path, chunk.type, or other conditions when needed.
6. Call knowhere_read_result_file for manifest.json, hierarchy.json, kb.csv, or table HTML when the answer depends on raw package data.

Response Style#

Keep tool-driven replies short and labeled.

• Reuse labels such as Scope, Source, File, Chunks, Job ID, and Next
• Prefer one short status line plus the key fields the user needs for the next step
• Keep path in your reasoning and answers when possible
• Cite chunkId and path when answering from retrieved chunks

Retrieval Rules#

• Prefer knowhere_grep for text search
• Use knowhere_preview_document before broad reads when the document is large or the relevant branch is unclear
• For image or table questions, inspect matching image or table chunks and related manifest asset entries before answering
• Do not rely on full.md alone when the question depends on exact structure, tables, or images
• When a tool response contains truncatedStrings: true, retry with a higher maxStringChars before answering

Attachment Markers#

When a prompt contains a marker like:

text

[media attached: /absolute/path/to/file.pdf (application/pdf) | handbook.pdf]

Use:

• the exact absolute path as filePath
• the visible filename as fileName

Tool Usage Examples#

Ingest a local file:

json

{
  "filePath": "/tmp/uploads/handbook.pdf",
  "fileName": "handbook.pdf"
}

Ingest a URL:

json

{
  "url": "https://example.com/report-2026.pdf",
  "title": "Q1 Report"
}

Preview a document:

json

{
  "docId": "handbook-1234"
}

Search across chunk fields:

json

{
  "docId": "paper-pdf-a370ef58",
  "conditions": [{ "pattern": "npm audit" }]
}

Read manifest JSON:

json

{
  "docId": "handbook-1234",
  "filePath": "manifest.json",
  "mode": "json"
}