Tencent IMA Skill
Scanned@hyddd
npx machina-cli add skill @hyddd/tencent-ima-skill --openclawIMA Skill
Control the IMA (ima.copilot) desktop application for AI search and private knowledge retrieval.
Tools
ima_search
Launches IMA and performs a search. Supports "Private Knowledge Base" mode via special tags.
- query (required): The search query. Prefix with
@个人知识库or@knowledgeto search your private knowledge base (requiresconfig.json). - autoclose (optional): "true" to close the app after searching. Default: "false".
Implementation:
/usr/bin/python3 /opt/homebrew/lib/node_modules/clawdbot/skills/ima/scripts/ima.py "{query}" --autoclose="{autoclose}"
Configuration
To enable private knowledge base search, you must providing your knowledge_id.
The script looks for config in:
~/.clawd_ima_config.jsonskills/ima/config.json
Format:
{
"knowledge_id": "your_id_string"
}
Examples
- Public:
clawdbot ima_search query="DeepSeek analysis" - Private:
clawdbot ima_search query="@knowledge project update"
Overview
IMA Skill lets you control the IMA (ima.copilot) desktop app to run AI searches and retrieve private knowledge using a configurable knowledge_id. It uses the ima_search tool and supports a Private Knowledge Base mode by tagging queries with @个人知识库 or @knowledge (requires config.json).
How This Skill Works
You invoke ima_search, which launches the IMA app and runs a search via a Python script. If you prefix the query with @个人知识库 or @knowledge, it performs a Private Knowledge Base search, provided knowledge_id is configured in ~/.clawd_ima_config.json or skills/ima/config.json.
When to Use It
- You need to search public AI results in IMA for general topics (no private data).
- You want to query your private knowledge base stored in IMA's Private Knowledge Base mode.
- You want the IMA app to close automatically after a search (autoclose enabled).
- You need to switch between public and private searches using the @knowledge or @个人知识库 prefixes.
- You have to provide a knowledge_id in the config.json to enable private KB searches.
Quick Start
- Step 1: Ensure knowledge_id is configured in ~/.clawd_ima_config.json or skills/ima/config.json.
- Step 2: Run a search, e.g., clawdbot ima_search query="DeepSeek analysis" or clawdbot ima_search query="@knowledge project update".
- Step 3: (Optional) Set autoclose to true to close the IMA app after the search.
Best Practices
- Create and verify knowledge_id in ~/.clawd_ima_config.json or skills/ima/config.json before attempting private KB searches.
- Prefix private KB queries with @personal知识库 or @knowledge to ensure the correct mode is used.
- Test queries without autoclose first to verify results before enabling autoclose.
- Use the full path for the Python script to avoid environment path issues (e.g., /usr/bin/python3 ... ima.py).
- Keep config.json and knowledge_id secure and never share sensitive IDs publicly.
Example Use Cases
- Public: clawdbot ima_search query="DeepSeek analysis"
- Private: clawdbot ima_search query="@knowledge project update"
- Private with Chinese tag: clawdbot ima_search query="@个人知识库 2024年需求"
- Public with autoclose: clawdbot ima_search query="Quarterly report" --autoclose="true"
- Private with config: clawdbot ima_search query="@knowledge product roadmap"
