eidolon-search
AI Agent memory search using SQLite FTS5. 90%+ token reduction (10x+) compared to reading full files. Use when the agent needs to search through markdown memory files, daily notes, or any text corpus efficiently. Triggers on memory search, file search, knowledge retrieval, or when context window is limited and full-file reading is too expensive.
Why use this skill?
Optimize your AI agent memory with eidolon-search. Index markdown files for 90%+ token savings and lightning-fast keyword retrieval using SQLite FTS5.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/dev-jslee/eidolon-searchWhat This Skill Does
The eidolon-search skill provides a high-performance, lightweight full-text search capability for AI agents using SQLite FTS5 technology. By indexing local markdown documentation, knowledge bases, or daily notes into a specialized search database, it enables agents to retrieve relevant information snippets without the need to read entire files. This architecture results in a 90%+ reduction in token usage compared to traditional document-parsing methods, making it ideal for maintaining long-term memory in resource-constrained environments. By shifting the retrieval process to a structured indexed query format, agents can quickly isolate specific information across large, multi-file workspaces.
Installation
To integrate this skill into your environment, use the OpenClaw management utility. Run the following command in your terminal: clawhub install openclaw/skills/skills/dev-jslee/eidolon-search. Once installed, ensure you have initialized your index by running python3 scripts/build-index.py <memory_dir> <db_path>. This will create the memory.db file, which serves as the backbone for all future search operations. For subsequent updates to your documentation, simply re-run the build script to update the index.
Use Cases
This skill is best utilized in scenarios involving extensive documentation or personal knowledge management. It is particularly effective for agents tasked with maintaining project history, debugging large codebases where design docs are stored in markdown, or managing daily developer logs. It excels when the context window is limited, as the agent can perform multiple targeted queries to extract only the necessary context rather than flooding the prompt with raw text files. It is also perfect for agents that need to recall specific architecture decisions made months prior without having to search through every individual file manually.
Example Prompts
- "Search my notes for 'Physical AI roadmap' and provide the top 5 relevant snippets to help me draft the current status update."
- "I need to know the specific technical requirements for the Jetson Orin integration; please search the memory database."
- "Look through the project timeline notes and tell me what the scheduled milestones are for Q3 using eidolon-search."
Tips & Limitations
Because this tool relies on FTS5 (keyword-based indexing), it does not perform semantic or vector-based search. It works best with precise, domain-specific terminology. To optimize results, use quotes for exact phrases, leverage boolean OR operators for synonym grouping, and keep your queries focused. Avoid using this tool for single, small files where direct reading is more efficient, or for queries requiring abstract, conceptual understanding rather than keyword presence.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-dev-jslee-eidolon-search": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-read, code-execution