2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill cheese-brain- _meta.json277 B
- SKILL.md8.3 KB
Overview
This skill is a DuckDB-powered knowledge management system for fast retrieval across 22+ entity types (projects, contacts, tools, workflows, decisions, etc.). It provides persistent, searchable memory with sub-millisecond keyword lookups and BM25 full‑text search for relevance-ranked discovery. Use it to recall project context, find configuration details, surface contact info, and retrieve workflows or past decisions instantly.
How this skill works
The skill stores entities as JSON-backed records (id, title, category, tags, data, timestamps) in a DuckDB store and exposes fast CLI queries. It supports two search modes: ultra-fast keyword search for exact/quick lookups and BM25 full-text search (FTS) for relevance-ranked discovery across many entities. You can add, update, list, export, and restore entities; sensitive fields are redacted by default and encrypted export is supported.
When to use it
- Recall project details, repo paths, schedules, and run commands quickly
- Find tools, scripts, or workflow steps and their locations/usage
- Retrieve contact information, calendar feeds, timezones, and channel IDs
- Search past decisions, rationale, and dates for context or audits
- Discover procedures and step‑by‑step workflows when troubleshooting or restoring systems
- Export, backup, or restore the entire knowledge base for disaster recovery
Best practices
- Tag consistently using lowercase, hyphenated tags (e.g., email-monitoring)
- Store structured fields (paths, URLs, schedules) in the JSON data field for reliable retrieval
- Search first to find candidate entities, then use get <id> to fetch full details
- Use FTS for discovery in large bases (>100 entities); use keyword search for quick exact matches
- Soft-delete rather than hard-delete so entries remain recoverable
- Avoid storing secrets in plain text; reference password manager locations and use encrypted exports
Example use cases
- "What’s the email monitor project?" → instant project entity with repo, path, schedule, run command
- "Where’s the backup script?" → tool entity with path, usage notes, related workflows
- "Calendar feed URL?" → contact entity returns ICS feed URL and timezone
- "How do I restore config?" → workflow entity provides step-by-step restore procedure
- Daily automated backup: export to parquet/json and store encrypted for offsite recovery
FAQ
Keyword searches return in under 1ms; FTS queries typically complete in ~5ms depending on DB size.
How do I protect secrets stored in data?
Sensitive keys are redacted by default. Don’t store plain-text secrets—reference a password manager (1Password, Bitwarden) and use encrypted export when backing up.
When should I create the FTS index?
Create the FTS index once you have a larger knowledge base (>100 entities) or when multi-word relevance ranking improves results.